A medium originally developed for differential enumeration of probiotic species in fermented milk (mMRS- BPB, Lee and Lee 2008) was evaluated for its ability to correctly discriminate members of the Lactobacillus plantarum (L. plantarum) and L. casei groups from other species. The medium was tested on 461 strains of lactic acid bacteria (LAB) belonging to eight genera and thirty-five spe- cies. Colony morphology was relatively consistent for L. plantarum, L. paraplantarum, L. pentosus, L. paracasei, L. casei, and L. rhamnosus, but, when used alone, was not always sufficient to discriminate these species from other species potentially present in cheese. A procedure based on tree classification was developed to obtain preliminary iden- tification on the basis of colony morphology, cell morphology, and CO2 production from glucose. By combining results of the tree classification procedure and heuristic rules, correct preliminary identification at the species or group level could be obtained in 74.4 % of cases overall, and the percentage of correct identifications was as high as 88-100 % for members of the L. plantarum and L. casei groups. When species belonging to groups that can be easily discriminated by rapid molecular methods were combined, the decision tree allowed to correct identification at the group level for the 95 % of the strains. Logistic regression was used to evaluate the effect of strain, operator, light source, and incubation temperature. Although all factors significantly affected one or more of the characters used for identification, the classification procedure proved to be quite robust. It may be difficult to use mMRS- BPB in the differential enumeration of LAB in cheese, except when species composition is relatively simple; however, it can be used as a simple tool to guide molecular identification in studies focused on the isolation of new strains from cheese.