It has been assumed that penicillin (and also other cell wall inhibitors) kill pneumococci predominantly by triggering their major autolytic enzyme (an N-acetylmuramoyl-L-alanine amidase; referred to as amidase), resulting in massive cell wall degradation. Three types of experiments suggest that only part of this killing is due to cell lysis by amidase. (i) Suppression of penicillin-induced lysis by specific inhibitors of amidase protected pneumococci only marginally from killing in spite of prolonged exposure to concentrations of penicillin that were 10x, 20x, or 100x greater than the MIC. (ii) Mutants from which the amidase was completely eliminated by plasmid insertion or deletion (Lyt-) were still killed, albeit at a slower rate than the wild-type Lyt+ strains (3 to 4 log units instead of 4 to 5 log units per 6 h, i.e., about 1 log unit slower than the wild type; P less than 0.001). (iii) A new mutation (cid), which was not related to the amidase gene, further reduced killing of mutants lacking amidase to 1 log unit per 6 h (Lyt- Cid- phenotype). Reintroduction of the amidase gene into Lyt- Cid- cells partially restored penicillin-induced lysis but increased only slightly the rate of killing (from 1 log unit per 6 h in Lyt- Cid- cells to 2 log units per 6 h in Lyt+ Cid- cells). We conclude that penicillin kills pneumococci by two distinct mechanisms: one that involves the triggering of the amidase (about 1 log unit of killing per 6 h) and another, amidase-independent mechanism that is responsible for 3 to 4 log units of killing per 6 h. Triggering of the amidase activity in situ in growing bacteria was significantly reduced in Lyt+ Cid- cells, indicating that there is some regulatory interaction between the cid gene product and the amidase.