The UK shellfish industry has recently been affected by the statutory closure of several cockle beds, following the detection of samples causing rapid and severe reactions in the regulatory approved test for diarrhetic shellfish poisoning (DSP) toxins, the mouse bioassay (MBA). It is contended that these socalled atypical results are due to procedural artefacts of the MBA; so far, several studies have failed to identify their cause. This paper critically assesses the development, regulatory use and methodological deficiencies of the MBA. It also discusses how testing for DSP toxins could and should have been improved and made more humane by applying the Three Rs concept of Reduction, Refinement and Replacement, and by the proper validation of the test method used. It is concluded that the MBA should not have been developed for the routine screening of shellfish samples, as it has a substantially severe endpoint and is not used as part of a tiered-testing strategy with non-animal methods. Moreover, during the UK monitoring programme for DSP toxins, the assay has been used without an optimised and universal protocol, and apparently without due regard to the principles of basic scientific methodology. In view of this, the atypical results obtained for cockle samples cannot be relied on as evidence of a human health