Combinations of gentamicin with cefotaxime, moxalactam, and ceftazidime were tested against 43 bacterial strains, most of them blood isolates. With an interaction index of 0.5 as borderline, synergism was demonstrated against 30 to 40% of the strains by the fractional inhibitory concentration index and against 50 to 70% by the fractional bactericidal concentration index. The reproducibility of the index was within ±0.2 for two-thirds of 40 repetitive assays and within ±0.4 to 0.5 for all of these assays. Similar results were obtained when netilmicin was substituted for gentamicin. The killing curve system for studying antibiotic synergism was standardized to give results comparable to those obtained with the interaction index. This was achieved when one-half of a previously determined minimum bactericidal concentration was used for single drugs and the amount of antibiotic was at least halved again when drugs were used in combination. An initial bacterial concentration of 105 to 106 colony-forming units per ml is recommended. Given these conditions, synergism could be defined as a 2-log10 or more decrease in viable count given by both drugs together, as compared with the more active of the pair after 24 h. Prediction of killing curve results could then be obtained with the fractional bactericidal concentration index. When cephalosporins and gentamicin were combined from the start, the beta-lactam antibiotics were less susceptible to inactivation, as demonstrated in time-killing assays. If one of the antibiotics were added after 24 h, synergism was not demonstrable. The results indicate that the new cephalosporins may be advantageously combined with aminoglycosides.