Hyperplastic nodules and hepatocellular carcinomas were induced in livers of rats by a low-protein diet containing 0.05% of the carcinogen N-2-fluorenylacetamide. Ganglioside amounts and composition were determined for histologically different hepatocellular carcinomas and compared with those for control livers, hyperplastic nodules, and liver tissue surrounding hepatomas and nodules as well as those for livers of fetal, newborn, 1-week-old, weanling, and adult Sprague-Dawley rats. Ganglioside sialic acid levels were elevated above those of normal adult liver in all liver tissues following the carcinogen treatment regimen. Livers of fetal and newborn rats contained nearly twice the amount of ganglioside sialic acid on a protein or DNA basis as did livers of adult rats. Analyses of individual nodules and hepatomas revealed two populations of tumors in which the levels of ganglioside sialic acid were 2.3 and 3.8 times normal. Ganglioside sialic acid content was at hepatoma levels in small nodules. Individual gangliosides were evenly distributed between products of the monosialoganglioside and disialoganglioside pathways in normal liver with a ratio of [N-acetylneuraminic acid (sialic acid)] (NAN)-galactose (Gal)-N-acetylgalactosamine (GalNAc)-(NAN)-Gal-glucose (Glc)-ceramide (Cer) (GD1a) to Gal-GalNAc-(NAN)2-Gal-Glc-Cer (GD1b) of about one. In contrast, the monosialogangliosides predominated in liver tissues following administration of the carcinogen. Increased levels of specific monosialogangliosides were present in nodules, in liver of carcinogen-treated animals prior to the appearance of tumors, and in the liver tissues surrounding nodules and hepatomas. In single hepatomas, ganglioside patterns correlated with tumorigenicity. A well-differentiated hepatoma had a normal complement of most gangliosides but was deficient in trisialogangliosides. In a poorly diferentiated but well-circumscribed hepatoma, the relative levels of all higher gangliosides were reduced. The monosialoganglioside Gal-GalNAc-(NAN)-Gal-Glc-Cer (GM1) accounted for 80% of the total ganglioside in a poorly circumscribed and poorly differentiated hepatoma. The ganglioside pattern of fetal livers most closely resembled that of a poorly differentiated hepatoma. During the first week post natum, levels of all higher monosialogangliosides and disialogangliosides declined, but the decline was most pronounced for gangliosides GM1 and GD1a. The ratio of GM1 + GD1a to GD1b + NAN-Gal-GalNAc-(NAN)2-Gal-Glc-Cer or (NAN)3-Gal-Glc-Cer (GT), used as an index of the relative predominance of the monoslaloganglioside and disialoganglioside pathways, fell from 2.7 for fetal liver to 0.4 for adult liver. Pools of precursor gangliosides increased during development, transiently for GalNAc-(NAN)-Gal-Glc-Cer and for more than 3 weeks for NAN-Gal-Glc-Cer. When hyperplastic nodules and hepatocellular carcinomas were compared, a reverse pattern was observed. The ratio of GM1 + GD1a to GD1b + GT rose steadily to values of 2.7 and 11...