A direct bacterial count method applicable to fish muscle containing over approximately 1 × 105bacteria per gram, with a modification employing preliminary incubation for muscle containing fewer bacteria, is described. Direct counts generally yielded similar and somewhat higher results than viable counts. Mixed bacterial populations of fish muscle generally exhibited the growth phases characteristic of a pure bacterial culture. There was absolutely no relation between direct (or viable) counts obtained subsequent to such incubation and the initial viable counts of the samples studied. Bacteria increased much more rapidly in excised fish muscle than in whole fish, and somewhat more rapidly in halibut than in salmon flesh.