Epoxy floors peel when the weakest plane in the substrate-and-coating assembly can no longer carry the stresses acting on it. The visible flap or chip does not identify that plane. A coating can separate from concrete, pull weak concrete apart, split between coats, or fracture within a layer—and each failure needs a different response.

Do not recoat until you know what failed. Preserve a representative sample, inspect both sides of it and the exposed floor, map the extent, review installation records, and test plausible causes. Moisture is one possibility, not the default diagnosis.

Peeling, flaking, and delamination are related search terms used inconsistently in the field. This guide focuses on loss of attachment or separation. Raised bubbles, pinholes, and blisters have additional mechanisms covered in the separate bubble, blister, and pinhole diagnostic.

Start with the failure plane

ASTM D7234 describes pull-off testing for coatings on concrete. Its public scope explains that failure occurs along the weakest plane among the loading adhesive, coating layers, and substrate, exposing that plane at the fracture surface. ICRI likewise distinguishes bond failure and cohesive rupture in its concrete repair terminology.

That is the right model even before laboratory testing: where is material missing, and what remains on each side?

Failure pattern What the fracture surfaces may show Questions it raises
Adhesive failure at concrete/coating interface Relatively clean concrete beneath; underside of chip mostly coating Was the surface profiled, clean, dry enough, and compatible with the primer or base coat?
Cohesive failure in concrete Cement paste or aggregate attached to underside of coating; fresh concrete fractured below Was the near-surface concrete weak, dusty, laitant, scaled, or damaged? Was removal too aggressive?
Intercoat failure One coating layer remains firmly attached while another separates Was the recoat window missed? Was the lower coat contaminated, glossy, undercured, or chemically incompatible?
Cohesive failure within coating Layer splits within its thickness rather than at an interface Was the mix, cure, film build, or service exposure outside the product limits?
Mechanical damage Gouges, impact marks, cuts, abrasion paths, or edge loading rather than spontaneous lifting Did traffic, tools, tires, chemicals, or impact exceed the installed system’s design?
Mixed failure Different materials remain across one sample or different zones Are multiple mechanisms present or does substrate quality vary?

This table is a screening framework, not a remote diagnosis. Dirt on a sample, damage caused while lifting it, or a repair attempt can obscure the original fracture surface.

Preserve evidence before removal

Before grinding away the floor:

  1. Photograph wide views and close-ups with locations and scale.
  2. Map isolated, linear, edge, traffic-lane, doorway, drain, and widespread areas.
  3. Note when the failure appeared and whether it is progressing.
  4. Save intact chips in labeled bags; photograph both faces.
  5. Record whether exposed concrete is clean, dusty, damp, dark, oily, smooth, or visibly fractured.
  6. Tap or sound adjacent areas where a qualified procedure supports it.
  7. Gather product names, batch numbers, TDS/SDS revisions, preparation records, moisture results, temperatures, dew point, mix times, coverage, film measurements, recoat times, and cure-to-service dates.
  8. Record cleaning chemicals, spills, tire traffic, heat, UV, water, impact, and changes in use.

Widespread or consequential failures justify independent coating, concrete, moisture, or engineering expertise. A manufacturer representative may help evaluate its own system, but disputed failures benefit from neutral evidence collection.

Cause family 1: insufficient profile or incomplete removal

A smooth, hard-troweled surface or an incompletely removed sealer can leave too little mechanical texture or an incompatible bond plane. The guide to epoxy over paint, sealer, or old epoxy explains why a new floor inherits the risk of any retained layer. Merely seeing grinder scratches does not prove that the coating’s required CSP was achieved. Conversely, an overly aggressive method can expose or create weak near-surface material.

Review the system’s required profile and accepted method, then compare field and edge areas. Look for:

  • smooth concrete on the exposed floor;
  • patches of old coating, sealer, adhesive, curing compound, or laitance;
  • different preparation at walls, corners, columns, and repairs;
  • stripes or missed passes;
  • polished zones caused by glazed or unsuitable tooling;
  • dust or loose aggregate trapped at the interface.

Our guide to grinding concrete for epoxy explains why removal tooling and final profiling are separate objectives.

Cause family 2: contamination

Oil, grease, silicone, tire dressing, process chemicals, cleaner residue, dust, salts, and old maintenance products can interfere with wetting or adhesion. Grinding can spread a contaminant if decontamination was not completed first.

Evidence may include localized failure around parking positions or equipment, beading, discoloration, odor, residue on the concrete, or poor adhesion in test areas. None of these observations identifies the chemical by itself. Laboratory analysis or specialist decontamination may be needed for deep or industrial contamination.

A primer is not a universal contamination blocker. The substrate must meet the concrete preparation requirements of the selected assembly.

Cause family 3: weak or laitant concrete

If cement paste or aggregate remains attached to the coating underside, the bond may have exceeded the tensile strength of the near-surface concrete. That is not the same as clean-interface adhesion failure.

Potential contributors include laitance, dusting, scaling, poor consolidation, freeze-thaw damage, fire or chemical damage, overworked surface paste, weak toppings, incompatible repairs, and other deterioration. The scope and cause of weak concrete need evaluation before recoating.

ASTM D7234 notes that pull-off results and failure mode are important performance information. Its public scope also cautions that the test can fail within the coating, glue, or substrate. ASTM C1583 or ICRI 210.3R methods may be more appropriate when near-surface concrete or repair bond is the target. These are destructive, controlled tests and require the current standard or guideline—not a casual field improvisation.

Moisture can contribute to disbondment, blistering, or chemical and ionic effects beneath a low-permeability floor. Sources can include residual construction water, vapor movement, a missing or ineffective below-slab vapor retarder, groundwater, drainage, leaks, wet processes, and condensation.

But peeling does not prove moisture. A valid investigation asks:

  • Was standardized testing performed before installation, by which method, and under what conditions?
  • What limit did the exact coating assembly publish at that time?
  • Are there current patterns, efflorescence, darkening, corrosion, or moisture gradients?
  • Is water entering at a wall, joint, drain, crack, or plumbing source?
  • Did failure occur at the concrete interface, within weak concrete, or between coats?
  • Are soluble contamination or other mechanisms plausible?

Current testing describes current conditions. It may support a diagnosis, but it does not automatically reconstruct the slab condition on installation day. See concrete moisture testing before epoxy for the difference among in-situ RH, MVER, and comparative meter screening.

Do not install a moisture-control primer merely because a floor peeled. First establish that the product is appropriate for the actual source, slab, test results, preparation, and replacement system.

Cause family 5: intercoat failure and missed recoat windows

If the primer or base coat remains while the layer above peels, focus on that interface. Possible causes include:

  • exceeding the maximum recoat window without the specified abrasion or treatment;
  • applying too soon over an undercured or solvent/water-retaining layer;
  • amine blush, condensation, dust, overspray, oil, or site contamination between coats;
  • a glossy, unabraded existing coating;
  • incompatible products or an undocumented cross-brand assembly;
  • incorrect broadcast aggregate, incomplete removal of loose aggregate, or poor saturation;
  • a lower coat that was itself mixed or cured incorrectly.

Sherwin-Williams identifies a missed recoat window as one mechanism for intercoat delamination and explains that the window is affected by time and temperature. The actual window and recovery method remain product-specific.

The epoxy floor primer guide explains why primer and topcoat compatibility includes film condition, recoat time, and the manufacturer-designed assembly—not merely shared epoxy chemistry.

Cause family 6: mixing, application, and cure errors

Two-part coatings need their exact ratio, mixing method, batch size, pot life, film build, and environment. Potential failure contributors include:

  • off-ratio components;
  • unmixed material from container sides or bottom;
  • incomplete pre-mixing where required;
  • extending material beyond pot life;
  • unauthorized thinning;
  • film that is too thin, too thick, or uneven;
  • substrate or material outside temperature limits;
  • condensation because slab temperature was too close to dew point;
  • water or chemical exposure before cure;
  • traffic before the documented return-to-service time.

Inspect retained material, batch records, coverage, tacky or soft areas, color differences, and the spatial pattern. Cure time and recoat time are not universal epoxy properties; use the exact historical TDS where possible. The cure and recoat guide separates working time, recoat windows, traffic stages, and full cure so the record is compared with the correct milestone.

Cause family 7: incompatible layers or service mismatch

A coating may adhere initially but fail when the assembly encounters heat, chemicals, UV, tire forces, impact, abrasion, or movement outside its design.

Examples include:

  • a rigid coating over an active crack or moving joint;
  • a topcoat not compatible with the cured base;
  • chemical exposure that softens or swells a layer;
  • outdoor UV or temperature cycling outside the system’s rating;
  • traffic or point loading beyond the specified build;
  • thermal shock or cleaning that exceeds the product limits;
  • snowplows, dragged equipment, sharp impact, or aggressive abrasion.

What about hot-tire pickup?

“Hot-tire pickup” is often used for coating that lifts beneath vehicle tires. Tire heat, plasticizers, turning force, contamination, cure state, film build, concrete profile, and system suitability may all be relevant. The parking-location pattern alone does not establish one mechanism. Treat it as a service-pattern clue, then examine the fracture plane and installation evidence.

Mechanical gouging or abrasion should also be separated from true adhesion loss. Recommending a new primer for damage caused by a steel edge would miss the service-design problem.

Localized versus widespread failure

Extent changes the repair decision:

  • One impact or isolated contaminant spot: localized repair may be possible after the cause and sound boundary are established.
  • Repeating parking spots or traffic lanes: investigate service, contamination, cure, and system suitability—not just patch adhesion.
  • Edges and doorways: compare preparation, moisture entry, termination details, UV, temperature, and traffic.
  • Regular stripes or machine paths: inspect preparation consistency and application technique.
  • Widespread hollow or peeling areas: removal to a verified sound substrate and complete system reassessment may be safer than spot repair.
  • Failure at joints or cracks: determine whether movement or water entry is active before selecting a detail.

Do not define the repair perimeter only by what can be lifted today. Adhesion testing, sounding, or exploratory removal may show that adjacent material is also compromised.

A diagnostic decision sequence

1. Locate the weakest plane

Inspect representative fracture surfaces. If uncertainty matters, use qualified pull-off or laboratory evaluation.

2. Map the pattern

Relate failure to edges, joints, traffic, drains, sunlight, contamination sources, equipment paths, and separate installation batches.

3. Reconstruct installation

Compare preparation, moisture evidence, environmental readings, mixing, coverage, recoat timing, and cure-to-service with the exact product documents.

4. Test competing hypotheses

Use moisture tests for moisture questions, contamination analysis for chemical questions, concrete tests for substrate strength, and adhesion evaluation for interfaces. One test should not answer a different question.

5. Define removal boundaries and replacement system

Remove unsound layers and substrate as required. Correct the underlying cause, reproduce the specified profile, and confirm compatibility from substrate through topcoat.

6. Prove the repair where consequences justify it

Manufacturer-approved mockups, test patches, or pull-off tests can verify the proposed method. Define acceptance before performing the test.

Why recoating over failure is risky

A new coat distributes resin over the surface; it does not restore the bond beneath a loose layer. If old coating remains, the new system is still attached through that old interface. Likewise, a moisture-control layer cannot repair active leakage, and a primer cannot strengthen every weak slab.

Benjamin Moore’s Corotech troubleshooting guide and Sika’s traffic-membrane manual both direct removal and renewed preparation for delaminated areas, with the extent depending on the failure. Those are manufacturer instructions for their product contexts, not a universal repair specification. The project still needs a compatible, documented replacement assembly.

Tools and products belong after diagnosis

This page can support future commercial guidance without letting products drive the conclusion:

  • pull-off or inspection equipment for a defined test plan;
  • RH probes, MVER kits, or comparative meters for the appropriate moisture question;
  • floor grinders, removal tooling, edge tools, and silica-controlled dust extraction;
  • compatible repair materials for the identified concrete defect;
  • the primer, base coat, aggregate, and topcoat within a documented replacement system;
  • environmental instruments and film-thickness tools for installation control.

The correct product category follows the established mechanism. Buying a stronger grinder before preserving a sample can destroy the most useful evidence.

When to bring in a professional

Seek specialist evaluation for widespread or recurrent failure, active water, structural movement, industrial chemical contamination, weak concrete over a large area, warranty disputes, occupied critical facilities, or any repair where downtime and service consequences are substantial.

The aim is not a more elaborate opinion. It is a traceable diagnosis supported by fracture surfaces, testing, records, and the selected system’s requirements.

Final rule: repair the cause and the interface

An epoxy flooring system is only as durable as its weakest relevant layer. Diagnose where separation occurred, determine why that plane was weak, correct the substrate or process, and verify the replacement assembly before recoating.

Similar visible peeling can result from different mechanisms. A technically credible repair begins by keeping those possibilities open until the evidence closes them.

ASTM D7234 and ICRI 210.3R are copyrighted, paid documents. This article uses their public titles, scopes, significance statements, and descriptions to explain the role of pull-off strength and failure mode. It does not claim full-standard access or provide a field procedure. Testing represented as compliant requires the complete current document and competent execution.

Sources and technical references

Accessed September 2026. Product documents can change; verify the current revision before specifying or installing a system.

  1. ASTM D7234-22: Pull-Off Strength of Coatings on Concrete — ASTM International
  2. 210.3R-2022: Guide for In-Situ Tensile Pull-Off Tests — International Concrete Repair Institute
  3. Concrete repair terminology — International Concrete Repair Institute
  4. Floor and wall coating failures — Sherwin-Williams High Performance Flooring
  5. Corotech floor coating preparation and troubleshooting guide — Benjamin Moore & Co.
  6. Traffic membrane troubleshooting manual — Sika USA
  7. Preventing concrete and epoxy floor blistering and disbondment — Sherwin-Williams High Performance Flooring