Ensuring the reliability and precision of assay results requires careful attention to assay design. In this case study we describe validation studies of an in vitro assay for botulinum neurotoxin type A based on its endopeptidase activity towards immobilised synthetic substrate. This assay, in common with many in vitro assays, is sensitive to changes in reagents and assay conditions and is time dependent. In addition, the toxin is not stable in solution. Differences in estimates of potency, resulting from positional factors, which are not significant in individual assays, are shown to be consistent and statistically significant over a longer series of assays. This study emphasizes that assay validation should not be viewed as a single step in assay development but must be considered as a continuing process if assay results are to be reliable and reproducible.