AIMS:To assess whether assimilation tests in isolation remain a valid method of identification of yeasts, when applied to a wide range of environmental and spoilage isolates. METHODS AND RESULTS:Seventy-one yeast strains were isolated from a soft drinks factory. These were identified using assimilation tests and by D1/D2 rDNA sequencing. When compared to sequencing, assimilation test identifications (MicroLog™) were 18·3% correct, a further 14·1% correct within the genus and 67·6% were incorrectly identified. The majority of the latter could be attributed to the rise in newly reported yeast species. CONCLUSIONS:Assimilation tests alone are unreliable as a universal means of yeast identification, because of numerous new species, variability of strains and increasing coincidence of assimilation profiles. Assimilation tests still have a useful role in the identification of common species, such as the majority of clinical isolates. SIGNIFICANCE AND IMPACT OF THE STUDY:It is probable, based on these results, that many yeast identifications reported in older literature are incorrect. This emphasizes the crucial need for accurate identification in present and future publications.