The in vitro metabolizing method was modified and its ability to correctly identify eight known hemolytic and nine known nonhemolytic drugs of glucose-6-phosphate (G6PD)-deficient erythrocytes was evaluated. The technique is based on inducing in vitro drug metabolism by incubation of red cells and drug with a reduced NADP-generating system in the presence of phenobarbital-induced mouse liver microsomes. Thus, this system provides a model for in vivo metabolic function. The hemolytic potential of tested drugs is indicated by the extent of loss of reduced glutathione of G6PD-deficient erythrocytes during 60-min incubations. Complete agreement between the test and literature for nonhemolytic compounds was observed. The test also correctly identified six of the eight known hemolytic drugs and failed to identify two known hemolytic drugs (acetanilide and sulfacetamide). The test was also applied to 14 drugs about which there is uncertainty regarding hemolytic potential. Of the latter, DL-alpha-methyldopa; alpha-naphthol; beta-naphthol; 2,3, dimercaptopropanol; phenacetin; and menadione were found to react positively. We conclude that this in vitro assay system will be useful in predicting which new drugs will be hemolytic in G6PD-deficient patients.