The possibility of using PCR for rapid identification of food-borne Staphylococcus aureus isolates was evaluated as an alternative to the API-Staph system. A total of 158 strains, 15 S. aureus, 12 other staphylococcal species, and 131 isolates recovered from 164 food samples were studied. They were phenotypically characterized by API-Staph profiles and tested for PCR amplification with specific primers directed to thermonuclease (nuc) and enterotoxin (sea to see) genes. Disagreement between the PCR results and API-Staph identification was further assessed by the analysis of randomly amplified polymorphic DNA (RAPD) profiles obtained with three universal primers (M13, T3, and T7) and 16S rDNA sequencing. Forty out of 131 isolates (31%) tested positive for PCR enterotoxin. Of these, 14 (11%) were positive for sea, 22 (17%) for sec, one (0.8%) for sed, and three (2.2%) for sea and sec. No amplification corresponding to seb nor see was obtained. Cluster analysis based on RAPD profiles revealed that most of the sec positive food isolates grouped together in three clusters. Cluster analysis combining the three RAPD fingerprints (M 13, T3, and T7), PCR-enterotoxin genotype and API-Staph profiles, grouped the nuc PCR positive isolates together with S. aureus reference strains and the nuc PCR negative isolates with reference strains of other staphylococcal species. The only nuc PCR positive food isolate that remained unclustered was a sed positive strain identified by 16S rDNA sequence as S. simulans. The high concordance between S. aureus and nuc PCR positive strains (99%) corroborates the specificity of the primers used and the suitability of nuc PCR for rapid identification of S. aureus in routine food analysis.