SUMMARY:Mice were challenged by mouth with a suspension containing equal numbers of streptomycin-sensitive (Str-) and streptomycin-resistant (Str+) variants of Salmonella typhimurium. These variants were of equal virulence but the Str+ variant grew more slowly in Vivo than the Str-variant. The LD50 dose contained c. 5 x lo6 bacteria. Heart blood obtained from mice dying from many LD50 doses nearly always contained a great excess of the S t r variant, but blood from mice dying from less than one LD 50 dose contained either Str-, Str+, or a mixture of Strand Str+ variants. The appearance of the Str+ variant alone in the latter mice strongly suggests that these fatal infections were initiated by a very small number of organisms or possibly by a single organism. It is therefore concluded that these organisms were acting independently. In this system, it is likely that any bacterium which enters the tissues from the gut can initiate a fatal infection and that the probability of effecting such an entrance almost entirely determines the probability of an inoculated bacterium causing a fatal infection.The inoculation of a partially resistant host with many bacteria will often cause a fatal infection when the inoculation of one bacterium is very unlikely to do so. Of several hypotheses advanced to account for this phenomenon, the most satisfactory appears to be that described elsewhere (Meynell & Stocker, 1957) as the 'hypothesis of independent action' which postulates ( a ) that bacteria act independently after inoculation, and (b) a mean probability (1 > p > 0) per inoculated bacterium of initiating a fatal infection which is constant and unaffected by the number of bacteria inoculated. On this hypothesis therefore, there will always be a chance that the death of a host inoculated with many bacteria will be caused by the progeny of only one of the inoculated bacteria. Although many of the inoculated bacteria may multiply to some extent, the total toxic effect of their clones will never be fatal and will be negligible compared to the toxic effect of the clone descended from only one of the inoculated bacteria. The progenitor of this clone may thus be said to have initiated the fatal infection.A well-known alternative hypothesis is that of the minimal lethal dose which postulates that if the dose exceeds a critical minimum size, death is inevitable; while if it is smaller than the minimum, the host is certain to survive. Thus, a lethal dose might be supposed to saturate the host defences so that death follows the multiplication either of all the inoculated bacteria or of those bacteria not needed for saturation. I n the latter case, a fatal infection could * Present address.