Quantitative delamination resistance data of fibre-reinforced polymer-matrix (FRP) composites for quasi-static or cyclic fatigue loads are determined under different loading modes and load rates, respectively. Such data find use, e.g., in FRP materials’ development, materials’ selection, assessment of durability, or structural design. Round robins during test development yielded repeatability and reproducibility (coefficients of variation) of roughly 10 to 25%. This scatter has several different sources. Intrinsic material variability amounts to a few percent at best, at least for advanced manufacturing processes. This intrinsic scatter is essential for material comparisons and structural design. Measurement resolution specified in standardised test procedures yields a maximum of 4–5% variability. Most of the remaining scatter comes from other, extrinsic sources. Test operator actions, e.g., choice of test set-up, manual data acquisition or data analysis can yield extrinsic scatter. Damage mechanisms during delamination initiation and propagation act on the micro- and meso-scale, typically a few micrometer to a few hundred micrometer in size, with corresponding time-scales estimated to between a few tens of nanoseconds and a few microseconds. Defect size-scales are hence several orders of magnitudes lower than test specimen and structural scales, respectively. Predictive capability of models using such test data for structures are, therefore, limited. Major issues are up-scaling from straight beam-like specimens to larger shell-like structures, possibly with complex shape and varying thickness as well as from unidirectional fibre orientation to multidirectional lay-ups.