Five laboratories participated in a joint ring study to investigate the role of bacterial cell number in the Salmonella mutagenicity test. A strictly standardized protocol, using sodium azide and TA 1535, was developed and employed to test the mutagenicity of N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) with different dilutions of Salmonella typhimurium TA-100 and TA-1535 cultures. All laboratories detected the mutagenic activity of sodium azide with only a 2-fold variation of test results. For MNNG the interlaboratory variation was approximately 5-fold. Decreasing numbers of test bacteria employed resulted in lower numbers of MNNG-induced revertants in all laboratories. The number of preexisting revertants decreased in direct proportion to the reduced cell content, whereas the number of spontaneous revertants was not as greatly affected. A critical amount of test bacteria was required in order to obtain numbers of induced revertants which were equal to twice the number of spontaneous revertants. Two evaluation parameters which may be employed to describe the mutagenicity of a compound are compared.