AIMS--To assess the quality of centres contributing antimicrobial susceptibility data to a centralised database. METHODS--Twelve organisms were distributed to 31 regional microbiology laboratories contributing data to a centralised susceptibility database. Participants were asked to determine susceptibilities to certain antibiotics by their routine method and return the data to the Department of Microbiology, Royal Hallamshire Hospital, Sheffield, for analysis. RESULTS--Results for the overwhelming majority of organism/antibiotic combinations were in agreement with expected results. Reasons for discrepancies included the non-bimodal distribution of susceptibilities, the use of different content discs, and, more importantly, minimum inhibitory concentrations falling close to breakpoint values. CONCLUSIONS--It is inevitable that any large multicentre database will contain a degree of inaccurate data. This study has highlighted several areas where discrepant results have occurred and has enabled Glaxo Laboratories to approach individual laboratories to address this problem. This study emphasises the value and consistency of Microbe Base as the largest database, of its kind, nationally.