Options were explored for fulfilling the legally required safety assessment for a widely applied epoxy/amine coating used for restoring corroded domestic drinking water supply systems. The coating was made up of two components mixed shortly before application, the first mainly consisting of bisphenol A diglycidyl ether (BADGE), the second of various amines. The analytically identified starting substances were all authorised, but only constituted a small proportion of the low molecular mass material left after curing and potentially migrating into water. Reaction products synthesised from constituents of the starting components (expected oligomers) could not be eluted from GC even after derivatisation, indicating that standard GC-MS screening would miss most potential migrants. They were detectable by size exclusion chromatography (SEC) after acetylation. HPLC with MS or fluorescence detection was possible for constituents including a BADGE moiety, but phenalkamines could not be detected with adequate sensitivity. Possibilities for determining long-term migration relevant for chronic toxicity are discussed. Analysis in water shortly after application of the coating overestimates migration if migration decreases over time and requires detection limits far out of reach. Analysis of a solvent extract of the coating is easier and provides an upper estimate of what could migrate into the drinking water over the years. However, to satisfy the regulatory requirements, components of the complex mixture need to be identified at lower proportions than those accessible. In vitro testing of the whole mixture for genotoxicity is expected to fail because of the required sensitivity and the glycidyl functions probably wrongly resulting in positive tests. The difficulties in dealing with this situation are discussed.