A numerical study of the fatty acid patterns of 263 reference strains belonging to the genera Arthrobacter, Aureobacterium, Brevibacterium, Cellulomonas, Clavibacter, Corynebacterium, Curtobacterium, Erysipelothrix, Microbacterium, and Rhodococcus was undertaken based on cultural and chemical standardized techniques. Clustering was by the unweighted pair group method using the correlation coefficient. Two cluster groups could be defined at the 62% level, one containing strains characterized by saturated and monounsaturated fatty acids and the second group characterized by iso- and anteiso-branched fatty acids. Within the first cluster group, a clear separation of strains assigned to the genera Rhodococcus, Erysipelothrix, and Corynebacterium could be achieved. Furthermore, strains of the species Corynebacterium glutamicum, Corynebacterium ammoniagenes, Corynebacterium diphtheriae, 'Corynebacterium ulcerans,' and Erysipelothrix rhusiopathiae could be found in distinct clusters, based on quantitative differences in fatty acid patterns. Within the second cluster group, a high degree of similarity between the genera Aureobacterium, Cellulomonas, Clavibacter, Curtobacterium, and Microbacterium found in phylogenetically based studies could be shown also by fatty acid patterns. Several strains of the plant pathogenic coryneform bacteria assigned to the genus Clavibacter and Curtobacterium flaccumfaciens were found within one cluster, indicating a high similarity between these genera. Strains of the genus Arthrobacter were grouped into three adjacent clusters and could not be differentiated by fatty acid patterns. The results of the study are essentially in line with a previously published numerical survey and with other chemotaxonomic and genetic data. Thus, quantitative fatty acid patterns are recommended for identification of several coryneform bacterial genera. In some cases an identification at the species level is possible.Key words: fatty acid analysis, coryneform bacteria, differentiation, numerical analysis.