1. The concentrations of penicillin G which (a) reduced the net rate of multiplication, (b) exerted a net bactericidal effect, and (c) killed the organisms at a maximal rate, have been defined for a total of 41 strains of α- and β-hemolytic streptococci, Staphylococcus aureus and Staphylococcus albus, Diplococcus pneumoniae, and the Reiter treponoma. 2. The concentration which killed the organisms at a maximal rate was 2 to 20 times the minimal effective level ("sensitivity" as ordinarily defined). With some organisms, even a 32,000-fold increase beyond this maximally effective level did not further increase the rate of its bactericidal effect. However, with approximately half the strains here studied (all 4 strains of group B β-hemolytic streptococci, 4 of 5 group C strains, 5 of 7 strains of Streptococcus fecalis, 2 of 4 other α-hemolytic streptococci, and 4 of 9 strains of staphylococci), when the concentration of penicillin was increased beyond that optimal level, the rate at which the organisms died was paradoxically reduced rather than increased, so that the maximal effect was obtained only within a relatively narrow optimal zone. 3. There were marked differences between bacterial species, and occasionally between different strains of the same species, not only with respect to the effective concentrations of penicillin, but also with respect to the maximal rate at which they could be killed by the drug in any concentration. Although there was a rough correlation between these two factors, there were many exceptions; individual strains affected only by high concentrations of penicillin might nevertheless be killed rapidly, while strains sensitive to minute concentrations might be killed only slowly. 4. Within the same bacterial suspension, individual organisms varied only to a minor degree with respect to the effective concentrations of penicillin. They varied strikingly, however, in their resistance to penicillin as measured by the times required to kill varying proportions of the cells.