top of page

ISO 4624
Pull-Off Test for Adhesion of Paints and Varnishes

Universal Grip ISO 4624 dolly alignment rack for pull-off adhesion tests

Downloads: ISO 4624 Dolly Alignment Rack for Pull-Off Adhesion Tests – Universal Grip Data Sheet (PDF)

Written by Chris Thomas, Universal Grip Company · Reviewed by Frank Bertini · Updated October 24, 2026

ISO 4624 is the international standard for the pull-off test for adhesion of paints, varnishes and related coatings. A steel or aluminium cylinder, the dolly, is glued to the cured coating and pulled off perpendicular to the surface at a controlled rate; the tensile stress at which the assembly breaks, and where it broke (in the substrate, at an interface, within a coat, in the glue or at the dolly), are the result. It is the quantitative complement to the cross-cut test: a cross-cut says pass or fail, a pull-off says 6 MPa with 60 % adhesive failure between primer and substrate, which is what a formulator or a protective-coatings inspector can act on.

This guide covers the current edition, ISO 4624:2023, and what changed from 2016; the three methods (A sandwich, B single dolly, C dolly as substrate) and which substrates each suits; dolly sizes from 7 to 100 mm; the adhesive rules including the 5 MPa glue-failure trigger; cutting around the dolly and the 150 µm exception; the ≤ 1 MPa/s rate with failure within 90 s; the strength formula with worked examples in newtons for every dolly size; the failure-code system from A to Y/Z with percentage area; portable testers versus a universal testing machine; the documented sources of error; a numeric comparison with ASTM D4541 and the concrete methods; and the dolly alignment rack and self-aligning pull-off couplings Universal Grip supplies. The ISO standards hub lists every ISO method we equip.

In this guide: What ISO 4624 Measures · Editions: 1978, 2002, 2016 and 2023 · Methods A, B and C: Which Substrate, How Many Dollies · Dollies, Centring Device and Cutting · Adhesives and Bonding · Test Conditions: Rate, Time, Conditioning and Replicates · Step-by-Step Procedure (Method A Sandwich on a Universal Testing Machine) · Breaking Strength Formula and Worked Examples · Failure Codes: Reading the Broken Faces · Portable Gauges vs a Universal Testing Machine · Sources of Error and Their Size · ISO 4624 vs ASTM D4541, D7234 and EN 1542 · Equipment for ISO 4624 from Universal Grip · FAQ

What ISO 4624 Measures

The standard applies to single- and multi-coat systems of paint, varnish and related products on any substrate. Dollies are bonded to the coating, and a tensile stress perpendicular to the plane of the coated substrate is increased at a uniform rate until something breaks. The result is the stress at the weakest point, which may be an adhesive failure between two layers (coating to substrate, coat to coat, coat to glue, glue to dolly) or a cohesive failure within a layer (within the substrate, within a coat, within the glue). The standard says plainly that the methods are most useful for relative ratings of panels whose adhesion differs significantly, because the measured stress depends on the dolly, adhesive, cutting and rate as well as on the coating. That is why every report carries the method letter, dolly diameter, adhesive, instrument type and failure codes alongside the number.

Related methods are kept distinct: ISO 2409 and ASTM D3359 are the cross-cut and tape tests for thin coatings; ISO 16276-1 applies the ISO 4624 pull-off to protective paint systems on steel in the field; ASTM D7234 and EN 1542 are the 50 mm dolly concrete methods; and ISO 4587 and ASTM D1002 measure the shear strength of structural adhesive joints, which is a different question from whether a paint stays on.

Editions: 1978, 2002, 2016 and 2023

ISO 4624 was first published in 1978, revised in 2002 (the edition that the Chinese GB/T 5210-2006 still adopts), in 2016 and again in 2023. The fourth edition, ISO 4624:2023, is the current one and is adopted in Europe as EN ISO 4624:2023; several vendor pages still cite 'EN ISO 4624:2019', which does not exist, or list the 1992-era number EN 24624 beside it. The 2023 foreword lists the changes:

  • Clause 3 terms added, so adhesive failure, cohesive failure and the layer notation are now defined in the standard rather than borrowed.

  • Dollies up to 100 mm diameter are allowed for concrete and similar substrates (5.2), aligning ISO 4624 with the 50 mm practice of ASTM D7234 and EN 1542 and the larger cores used on rough masonry.

  • A new sentence in Clause 6 on choosing the adhesive in relation to the substrate and coating.

  • More guidance for the single-dolly method (9.4.2), the field method on rigid substrates.

  • Normative references updated: ISO 1513 and ISO 1514 (test panels), ISO 2808 (film thickness), ISO 3270 (conditioning), ISO 4618 (terms) and ISO 15528 (sampling).

Methods A, B and C: Which Substrate, How Many Dollies

The three methods differ in how the coated piece is supported, and the choice is dictated by the substrate's stiffness. A thin steel panel, a plastic part or a wooden board bends under a single dolly's pull and peels the coating off at the edge rather than pulling it off uniformly; a thick steel plate or concrete block does not.

  • Method A, two dollies (the sandwich): the coated test piece is bonded between two coaxial dollies and the whole assembly is pulled apart in a tensile tester. It is the method for deformable substrates (thin metal, plastics, wood) and is also used on rigid ones. The assembly must be made in a centring device so the two dollies are coaxial and square; this is the Universal Grip alignment rack described below. A tensile machine with a self-aligning coupling is needed because the sandwich has to be gripped from both sides, which a portable gauge cannot do.

  • Method B, one dolly, one side only: a dolly is bonded to the coating on a rigid substrate, the coating is cut around it, and the dolly is pulled by a portable adhesion gauge or by a tensile machine fixture while the substrate is held. Rigid substrates only: thick metal plate, concrete, masonry.

  • Method C, dollies, one used as the painted substrate: the coating is applied to the face of a dolly, cured, and a second dolly is bonded on top; the pair is pulled apart. It removes the substrate variable and is used in formulation work and for comparing products on a standard steel surface; it also needs the centring device.

Common names for the methods, general, single-side and stud, appear on some pages, but reports should use the ISO wording and letter. In all three the stress is applied perpendicular to the coated surface and the alignment must be coaxial: a dolly pulled off-axis fails in peel at the high-stress edge and reads low, which is the single largest reason portable fixed-alignment gauges read about half of self-aligning ones.

Dollies, Centring Device and Cutting

  • Dolly material and size (5.2): steel or aluminium, nominal diameter 20 mm, face area 314 mm², faces machined perpendicular to the axis; a height of at least half the diameter is recommended (not required) so the dolly stays square under load. Smaller dollies than the 20 mm steel-substrate dolly may be used to reduce the force; 7 mm dollies are permitted for one-side tests where substrate distortion is a risk, with more replicates; up to 100 mm is allowed for concrete. The diameter is always recorded.

  • Dolly preparation: faces abraded (grit-blasted faces give measurably higher results than sanded ones in published trials) and degreased before bonding; a dolly base that is too thin flexes and lowers the reading.

  • Centring device (5.3): required for Methods A and C to hold the dollies coaxial while the adhesive cures. The standard's Figure 3 shows a jig with aligned bores; the Universal Grip rack is a two-plate version that holds eleven assemblies.

  • Cutting device (5.4): a sharp knife or hole-cutter that cuts through the adhesive and coating to the substrate around the dolly, so that only the coating under the dolly carries load. Cutting is the default; by agreement it may be omitted for coatings thinner than 150 µm. Whether the coating was cut, and with what tool, is reported. Published trials show cutting lowers the measured stress by roughly 9 % on a thick coating, which is the correct number, not an artefact.

Adhesives and Bonding

The adhesive must be stronger in cohesion and in bonding than the coating under test and must not visibly attack the coating; it is screened on a spare panel first. The standard names cyanoacrylates, two-component solventless epoxides and peroxide-catalysed polyesters as often suitable, and prefers a quick-drying two-pack epoxy in high humidity. The practical rule is in Clause 6: if the failure is mainly in the adhesive at stresses below 5 MPa, the adhesive is unsuitable and another is tried. Cyanoacrylates are quick and fine for smooth, strong coatings, but on porous or rough surfaces, and for sandwich assemblies that must survive handling, a two-part epoxy is the usual choice. Mixing matters: a published Elcometer trial found a wrong 1:1 mix of a 2:1 epoxy cut the measured adhesion by 41 %, from 6.7 to 3.9 MPa, while the coating had not changed at all. Cure the adhesive per the manufacturer, typically 24 hours at 23 °C, in the centring device for Methods A and C, and test in the standard atmosphere.

Test Conditions: Rate, Time, Conditioning and Replicates

The tensile tester (5.1) applies the stress perpendicular to the coated substrate at a substantially uniform rate not greater than 1 MPa/s, chosen so that the assembly fails within 90 s. Both limits matter: faster than 1 MPa/s inflates the result because coatings are viscoelastic, slower than 90 s to failure lets creep and peel initiation lower it. On a universal testing machine in force control the rate is set directly in newtons per second from the dolly area, 314 N/s for a 20 mm dolly at 1 MPa/s; a coating expected at 10 MPa then fails in about 10 s at the maximum rate and the rate must stay above 0.11 MPa/s (35 N/s) to finish inside 90 s. Other pull-off instruments are permitted if they give similar results, and the instrument type is part of the report. Panels are prepared per ISO 1514 with film thickness measured per ISO 2808, and conditioning and testing are done in the ISO 3270 standard atmosphere of 23 ± 2 °C and 50 ± 5 % RH; the standard calls for six determinations, so eleven positions in an alignment rack cover the minimum plus spares for the 5 to 10 % of pulls that fail in the glue.

Step-by-Step Procedure (Method A Sandwich on a Universal Testing Machine)

  • Cut coated test pieces slightly larger than the dolly, or use whole panels for Method B; measure the film thickness per ISO 2808.

  • Abrade and degrease the dolly faces; mix the adhesive accurately and apply a thin, void-free layer to both dollies.

  • Assemble lower dolly, coated piece and upper dolly in the centring rack, set the plate gap with the wing nuts so the bores hold both dollies coaxial, and leave to cure (typically 24 h at 23 °C) in the standard atmosphere.

  • Cut through adhesive squeeze-out and coating to the substrate around the dolly edge, unless the no-cut agreement for coatings under 150 µm applies; record the tool.

  • Mount the assembly between self-aligning couplings on the tensile machine so that no bending enters the stack; zero the force.

  • Pull in force control at ≤ 1 MPa/s (314 N/s for 20 mm) so that failure occurs within 90 s; record the maximum force.

  • Examine both faces, classify the failure by layer (A, A/B, B, B/C, C, −/Y, Y, Y/Z) and estimate each area to the nearest 10 %.

  • Repeat for six assemblies; calculate σ = 4F/(πd²) for each; report individual and mean values with the failure codes, method letter, dolly size, adhesive, cutting, instrument and rate.

Breaking Strength Formula and Worked Examples

The pull-off strength is the breaking force divided by the dolly face area: σ = 4F/(π·d²), with F in newtons and d in millimetres giving megapascals. For the nominal 20 mm dolly the area is π × 400/4 = 314.16 mm², so 1 MPa corresponds to 314 N and the conversion is simply force in newtons divided by 314. Worked examples:

  • 20 mm dolly, F = 1,880 N: σ = 4 × 1,880/(π × 400) = 7,520/1,256.6 = 5.98 ≈ 6.0 MPa. At the maximum rate of 1 MPa/s (314 N/s) this assembly fails in 6 s; a slower 0.3 MPa/s (94 N/s) gives 20 s and stays well inside the 90 s limit.

  • 5 MPa on 20 mm: F = 5 × 314.16 = 1,571 N, the glue-failure threshold below which a different adhesive must be tried.

  • 7 mm dolly (38.5 mm²): 1 MPa = 38.5 N; a 10 MPa coating fails at 385 N, which is why small dollies suit thin or deformable substrates and portable gauges.

  • 50 mm dolly (1,963 mm²): 1 MPa = 1,963 N; a 2.5 MPa concrete coating fails at 4.9 kN, so a 5 kN or 10 kN frame is needed even for weak substrates.

  • 100 mm dolly (7,854 mm²): 1 MPa = 7.85 kN; the new 2023 size for rough concrete puts a 3 MPa system at 23.6 kN, a 25 kN machine job.

  • Six-value report: 6.0, 5.6, 6.3, 5.9, 6.1 and 5.7 MPa give a mean of 5.93 ≈ 5.9 MPa; a seventh pull that failed at 4.2 MPa with 80 % Y (within the glue) is reported but flagged as a glue failure.

Units: 1 MPa = 1 N/mm² = 145 psi (not the 150 psi/s often printed for the rate) = 10.2 kgf/cm². A 20 mm dolly's 0.79 in diameter is sometimes mis-stated as 0.76 or 0.78 in on vendor pages; the area, not the diameter, is what the gauge scale is built on.

Failure Codes: Reading the Broken Faces

The number alone does not tell the formulator what to fix; the failure location does. ISO 4624 codes each layer and interface and records the fraction of the dolly face showing each type, to the nearest 10 %. The substrate is A, the first coat B, the second coat C and so on (n for a further coat), the adhesive is Y and the dolly is Z; a single letter means cohesive failure within that layer and a slash means adhesive failure between the two layers either side of it:

  • A: cohesive failure of the substrate, for example concrete or wood tearing below the coating; the coating is stronger than the substrate.

  • A/B: adhesive failure between substrate and first coat, the classic loss of primer adhesion.

  • B: cohesive failure within the first coat.

  • B/C: adhesive failure between first and second coat, an intercoat problem (contamination, overcoating window missed).

  • C: cohesive failure within the second coat; n/m generalises to further coats.

  • −/Y: adhesive failure between the final coat and the adhesive, usually a surface-preparation or adhesive-compatibility problem.

  • Y: cohesive failure within the adhesive; Y/Z: adhesive failure between adhesive and dolly. Both are glue failures: the coating survived at least that stress, so the result is a lower bound, and below 5 MPa the standard says to change adhesive.

A report line therefore reads, for example, '6 MPa, 60 % A/B, 40 % B'. The standard records every failure type; it does not declare Y 'invalid', as some vendor pages do; whether a glue failure counts is for the specification to decide (ASTM D4541 disregards results with 50 % or more glue failure unless they already exceed the required value).

Portable Gauges vs a Universal Testing Machine

ISO 4624 names a tensile tester as the instrument and allows other pull-off instruments if they give similar results. Portable gauges, the ASTM D4541 Types II to VI, are the field tools for Method B on rigid substrates: a mechanical fixed-alignment screw gauge, hydraulic and pneumatic self-aligning gauges and automatic models with rate control. Their limits are known: the fixed-alignment Type II reads about half of the self-aligning types because any misalignment turns the pull into a peel; manual gauges depend on the operator winding at a steady rate (about one turn per second approximates 1 MPa/s on common models); and none can grip a sandwich. A universal testing machine with a self-aligning coupling at each end runs all three methods, holds the rate to ±1 % in force control, records the full force-time curve so the 90 s rule and any slip are documented, and is the reference against which a portable gauge is validated for the 'similar results' clause. For a lab that qualifies coatings on thin panels or plastics (Method A) or develops formulations (Method C) the machine is the only route; for an inspector on a tank or bridge the gauge is, and the two should be correlated on the same coating before numbers are compared.

Sources of Error and Their Size

  • Adhesive mix ratio: a wrong 1:1 mix of a 2:1 epoxy lowered results 41 % (6.7 → 3.9 MPa) in published trials.

  • Dolly face preparation: grit-blasted faces gave 13.2 MPa against 10.8 MPa sanded, about 22 % higher, on the same coating.

  • Dolly base thickness: a 2.6 mm base read 8.8 MPa where a 4.0 mm base read 10.5 MPa, a 16 % difference from flexure alone; the standard's recommended height of at least half the diameter exists for this reason.

  • Cutting: scoring around the dolly lowered results about 9 % (10.4 → 9.5 MPa); it is the correct, specified procedure, so a no-cut result is not comparable with a cut one.

  • Cure temperature and time: 24 h at 22 °C gave the highest and most consistent results in the same trial; short or hot cures scatter.

  • Rate: faster than 1 MPa/s reads high, failure beyond 90 s reads low; report the rate and instrument.

  • Misalignment: off-axis pull converts tension to peel at one edge; fixed-alignment gauges read about half of self-aligning ones, and a tilted sandwich on a machine does the same.

  • Substrate distortion: a thin panel under a single dolly bends and fails at the edge; use Method A or a 7 mm dolly.

  • Wrong edition or formula: quoting the 2002 edition, 'EN ISO 4624:2019', σ = F/A without the dolly area, or 'Y is invalid' marks a report that will not survive audit.

ISO 4624 vs ASTM D4541, D7234 and EN 1542

  • Scope and instrument: ISO 4624 covers laboratory tensile testers and portable gauges, with three methods including the two-dolly sandwich; ASTM D4541 is written for portable testers only (Types II–VI), single dolly on a rigid or supported substrate.

  • Rate and time: ISO ≤ 1 MPa/s with failure within 90 s; D4541 ≤ 1 MPa/s (145 psi/s) with failure within 100 s.

  • Dolly: both nominally 20 mm; ISO allows 7 mm for one-side tests and up to 100 mm for concrete; D4541 uses 10, 14, 20 and 50 mm variants by gauge type.

  • Replicates: ISO six determinations; D4541 at least three per location.

  • Cutting: ISO cuts by default, no-cut by agreement below 150 µm; D4541 scores by agreement, recommended above 500 µm, and removed its old warning against scoring in 2017.

  • Glue failure: ISO reports Y and Y/Z with area and triggers an adhesive change below 5 MPa; D4541 disregards results with 50 % or more glue failure unless above the specified value.

  • Pass/fail option: D4541 Protocol 2 stops at a specified stress; ISO 4624 always tests to fracture.

  • Concrete: ASTM D7234 and EN 1542 use 50 mm dollies with the coating scored or cored into the substrate, at low rates suited to 0.3–4 MPa results; ISO 4624:2023 now reaches those substrates with its 100 mm dolly allowance.

  • Portable gauge bias: D4541's own precision data show Type II reading about half of self-aligning types and Type IV the most repeatable (r ≈ 15 %, R ≈ 28 %); ISO's 'similar results' clause makes the tensile machine the reference.

Equipment for ISO 4624 from Universal Grip

Dolly alignment rack (part UGC-ISO4624-TF). A centring and alignment rack for preparing pull-off adhesion test assemblies: two plates with matching bores hold test dollies coaxial and square while the adhesive cures, and wing nuts set the plate gap, so up to eleven assemblies are bonded with consistent alignment before being pulled apart on the test machine. Two drilled plates align upper and lower dollies coaxially during cure; eleven positions allow batch preparation; wing-nut adjustment sets plate spacing to suit assembly height; ground steel construction keeps the bores parallel and square; it is used with self-aligning pull-off fixtures on your universal test machine. Specifications: pull-off adhesion, sandwich (two-dolly) assemblies; function, alignment of dollies during adhesive curing; specimens, coated substrates bonded between test dollies; capacity, eleven assemblies per rack; material, stainless or corrosion-protected steel; bores sized to the customer's dolly diameter; wing-nut set plate spacing. It is the Clause 5.3 centring device for Methods A and C; it is a preparation jig, not a load-bearing fixture, and the eleven positions give the six-assembly minimum plus spares. Specify dolly diameter and height, number of positions and the pull-off fixture used when ordering.

Pull-off fixture and machine. The assembly is pulled between self-aligning couplings, universal-joint adapters that let each dolly find the load axis, mounted on a Mark-10 tensile tester with a force gauge for 7 to 20 mm dollies up to about 6 kN, or on the 2.5 kN or 5 kN benchtop UTM for 20 mm dollies on strong coatings (a 15 MPa system needs 4.7 kN) and 50 mm dollies on concrete coatings. Force-rate control sets the 314 N/s maximum for 20 mm dollies directly, and the recorded curve documents the 90 s rule. Environmental chambers allow pulls at service temperature. For the structural-adhesive questions that pull-off does not answer, shear strength and peel, see the adhesives testing page, the ISO 4587 lap-shear guide and the ASTM D1002 guide.

ISO 4624 FAQ

What is ISO 4624?

ISO 4624 is the international pull-off test for the adhesion of paints, varnishes and related coatings: a dolly is bonded to the coating and pulled off perpendicular to the surface at a uniform rate of not more than 1 MPa/s so that failure occurs within 90 s; the breaking stress in MPa and the location of failure are reported. The current edition is ISO 4624:2023 (fourth edition), adopted as EN ISO 4624:2023.

What are Methods A, B and C in ISO 4624?

Method A uses two dollies with the coated test piece bonded between them (the sandwich), for deformable substrates such as thin metal, plastics and wood as well as rigid ones; Method B uses one dolly on one side of a rigid substrate; Method C uses dollies, one of which is the painted substrate, for formulation work. Methods A and C need a centring device during cure and a tensile machine; Method B can be run with a portable gauge.

What dolly size does ISO 4624 use?

A steel or aluminium dolly of nominal 20 mm diameter (314 mm² face), with a height of at least half the diameter recommended; 7 mm dollies are permitted for one-side tests where substrate distortion is a risk, and since 2023 dollies up to 100 mm are allowed for concrete. The diameter used is always reported.

How is pull-off strength calculated?

σ = 4F/(π·d²), with F the breaking force in newtons and d the dolly diameter in millimetres, giving MPa. On a 20 mm dolly 1 MPa corresponds to 314 N, so 1,880 N gives 6.0 MPa; on a 50 mm dolly 1 MPa is 1,963 N and on a 7 mm dolly 38.5 N.

What rate does ISO 4624 require?

A substantially uniform stress rate not greater than 1 MPa/s, set so that the assembly fails within 90 s. On a universal testing machine in force control that is 314 N/s for a 20 mm dolly; to stay within 90 s the rate must exceed about 0.11 MPa/s for a 10 MPa coating. ASTM D4541 uses the same 1 MPa/s ceiling with a 100 s limit.

Do I have to cut around the dolly?

Yes by default: a sharp cutter is used through the adhesive and coating to the substrate so that only the coating under the dolly carries load; by agreement cutting may be omitted for coatings thinner than 150 µm. Cutting typically lowers the measured value by about 9 % and is the specified procedure, so cut and uncut results are not comparable; whether the coating was cut, and the tool, are reported.

Which adhesive should be used for the dollies?

One stronger than the coating that does not attack it: ISO 4624 lists cyanoacrylates, two-component solventless epoxides and peroxide-catalysed polyesters as often suitable and prefers a quick-drying two-pack epoxy in high humidity. If failure is mainly within the adhesive at stresses below 5 MPa, the adhesive is unsuitable and another is tried. Mix two-part adhesives exactly; a wrong mix ratio has been shown to cut results by 41 %.

How many tests does ISO 4624 require?

The standard calls for six determinations per coating system, reporting individual and mean values with their failure codes. Because 5 to 10 % of pulls fail in the glue, labs prepare spares; an eleven-position alignment rack covers the six plus spares in one cure.

What do the failure codes A, A/B, B and Y mean?

A is cohesive failure of the substrate, A/B adhesive failure between substrate and first coat, B cohesive failure within the first coat, B/C between first and second coat, C within the second coat, −/Y between the final coat and the adhesive, Y within the adhesive and Y/Z between adhesive and dolly; each is reported with its area to the nearest 10 %, for example '6 MPa, 60 % A/B, 40 % B'. Y and Y/Z are glue failures and give only a lower bound.

What is the difference between ISO 4624 and ASTM D4541?

Both pull a nominal 20 mm dolly at no more than 1 MPa/s, but ISO 4624 requires failure within 90 s, six determinations and a tensile tester or equivalent gauge with a two-dolly sandwich option for thin substrates, while D4541 is written for portable testers (Types II–VI), allows failure within 100 s, needs at least three tests, offers a pass/fail Protocol 2 and disregards results with 50 % or more glue failure. Fixed-alignment D4541 Type II gauges read about half of self-aligning instruments.

Can a universal testing machine be used for ISO 4624?

Yes; the standard's instrument is a tensile tester applying stress perpendicular to the coated surface, and other pull-off instruments are accepted only if they give similar results. A machine with self-aligning couplings runs all three methods, controls the rate in force control and records the curve; Method A sandwiches can only be pulled this way.

What machine capacity does a pull-off test need?

Dolly area times the expected strength: a 20 mm dolly at 15 MPa needs 4.7 kN, so a 5 kN benchtop frame or a Mark-10 tester covers most paint systems; a 50 mm dolly on a 2.5 MPa concrete coating needs 4.9 kN and a 100 mm dolly at 3 MPa needs 23.6 kN, which calls for a 25 kN frame.

Need a dolly alignment rack, a self-aligning pull-off fixture or a complete tensile system for ISO 4624? Contact Universal Grip with your dolly diameter, expected adhesion strength and machine connection and we will quote the right setup.

bottom of page