The methods and criteria for testing for synergy were compared with use of 22 strains of Klebsiella pneumoniae and a combination of cephalothin and gentamicin. With the checkerboard technique, the frequency of synergy varied depending on the criterion employed. When the rate of killing of bacteria by antibiotics was measured, results varied depending on the size of inoculum. When the checkerboard and killing-curve techniques were compared, there was poor correlation in terms of the frequency of strains showing synergy. Because clinical observations suggest that treatment of serious infections with combinations of antibiotics found to be synergistic in vitro may enhance survival, standardization of the methods and criteria used to test for synergy should be undertaken.