Multiple molecular forms of glucose-6-phosphate dehydrogenase (G6PD) in normal, preneoplastic, and neoplastic mammary tissues were separated by polyacrylamide gel electrophoresis and identified by specific straining for enzyme activity. Mammary tissue from lactating BALB/c mice showed considerable amounts (up to 50%) of a slower-migrating G6PD species, G6PD-III, which was essentially absent from glands of pregnant mice, preneoplastic nodules, and mammary carcinomas. All tissues possessed a faster-migrating species, G6PD-II, which accounted for up to 85% of the total G6PD in the glands of pregnant mice. A third species, G6PD-I, migrating more rapidly than G6PD-II, was found in both abnormal tissues (preneoplastic and neoplastic) and accounted for up to 35% of the total enzymatic activity. G6PD-I was present in moderate amounts (less than 15%) in glands from pregnant mice and was essentially absent from the lactating gland (approximately 5%). The addition of dithiothreitol did not alter the measurable G6PD activity but did increase the relative activity of G6PD-II or G6PD-I, as judged by the intensity of the bands on the gels. Mild oxidation (stirring overnight at 4 degrees in air) resulted in a loss of G6PD activity, but preparations had greater amounts of G6PD-III; presence of dithiothreitol during aeration partially prevented loss of G6PD activity and largely prvented the appearance of G6PD-III. Molecular-weight estimations with preparations from lactating mice yielded a value of 118,000 for G6PD-II and 260,000 for G6PD-III, suggesting a monomer and dimer, respectively. The addition of nicotinamide adenine dinucleotide phosphate stabilized G6PD activity by preventing heat inactivation at 47 degrees; nicotinamide adenine dinucleotide phosphate did not alter the pattern of species present. The data from heat inactivation studies suggest that G6PD-III (dimer) was the more stable species. The addition of nicotinamide adenine dinucleotide phosphate to samples after oxidation in the absence of dithiothreitol (about 70% loss of activity) resulted in no change in patterns and in recovery of full G6PD activity during heating at 47 degrees. A potential relationship between glutathione reductase activity and the pattern of G6PD species observed in the various tissues is noted.