Improvements in the reproducibility of nystatin agar diffusion assays have been achieved by the use of liquid nitrogen stored inocula and deep frozen standard stock solutions. The overall percentage variability of the assay has been reduced from over 5% with daily prepared standards and inocula to around 1 % with a frozen inocula and to 0.6% with a combination of frozen inocula and standards. The implications of these improvements in the standardization of nystatin assays, and microbiological assays generally are discussed.The variation in the results obtained from routine microbiological assay procedures among different laboratories using the same standardized supply of antibiotic is a cause for concern. The results for the international standard for nystatin highlight the variation and difficulties inherent in a daily preparation of both inoculum and standard (Lightbown et a1 1963). Whilst Arret & Eckert (1968) have stated that the 95 % confidence range of an average microbiological assay is 10 %, the British Pharmacopoeia (1973) requirement is ?~5 % and currently any result within this range is considered normal variation and any result beyond this range is considered significantly different. The main contributory causes of this variation in supposedly comparable assays are variations in the inoculum, in the method of preparation of standard solutions of the antibiotic and also in available media (for example, see Freeman et a1 1977).