The ATB 32C (bioMerieux, Spain), AMS-YBC (Vitek System, bioMerieux, Spain) and API 20C (bioMerieux, Spain) systems were evaluated for their reliability in identifying 100 clinical yeast isolates. The ATB 32C, AMS-YBC and API 20C systems correctly identified 97%, 98% and 100% of the isolates respectively. There were no significant differences in incubation periods between ATB 32C and AMS-YBC systems. One isolate of Candida tropicalis was wrongly identified by the ATB 32C and the AMS-YBC systems. The Saccharomyces cerevisiae isolate was wrongly identified by the ATB 32C system while the AMS-YBC failed to identify it and a third isolate of Candida krusei was wrongly identified by the ATB 32C system. The overall accuracy and rapidity of the ATB 32C and AMS-YBC systems were sufficient to permit recommendation of either of these systems for routine use in the clinical microbiology laboratory, although the first system enjoys the advantages of having a wider data-base and the possibility of manual reading.