StrMMARYThe systematic relationships among 54 strains of bacteria, representing principally the genera Achromobacter, Aerobacter, Alcaligenes, Eschmichia, Mima, Pseudomonas, Serratia and Streptococcus, were examined by computer methods. Seventy-one properties of these organisms were determined, and the resulting data scored in different ways (according to various proposed techniques) before being submitted to an appropriate computer program for calculation of similarity (S) values. These comparative studies indicated that better division of organisms into mutually similar groups can be achieved when data about properties which may have several alternative expressions are handled in the manner proposed by Beers & Lockhart (1962). The number of comparisons which contribute to individual similarity values should be held constant by adequate treatment of quantitative data and by adoption of scoring methods which permit comparisons between ' negative ' properties. It may be useful to employ distance (D = log,l/S) rather than similarity as the primary measure of relationships among groups of organisms.