A computer-assisted characterization of 296 Micrococcaceae isolates obtained from aerobically chill-stored lamb carcasses was carried out using a probability matrix and Bayesian identification theorems, complemented with cluster analysis. Preliminary identification was done with an original probability matrix comprising 37 previously described taxa and 32 tests. Although its statistical quality was adequate, the percentage of identification of field strains to species level was only 70% (96.6% identified with genera). To achieve an improved characterization, cluster analysis was subsequently performed on this group and an additional 26% could be associated with defined species, with five more taxa defined. The combined use of both approaches was judged positive as new identifications and better discrimination could be achieved. The majority of our isolates belonged to the Staphylococcus species group. Many species and groups of staphylococci increased as the spoilage progressed.