The association between hepatic lipidosis (HL) and disease in 59 anorectic, ketotic, lactating Holstein heifers and cows was investigated. Severe HL, as determined by histologic evaluation of liver tissue, was present in 46 animals; only half of these animals required intensive treatment for ketosis, and only half had serum biochemical evidence of liver disease, as determined by the presence of a last value of 2-fold or greater than the upper limit of the reference ranges for at least 2 of the 4 serum tests: gamma-glutamyl transferase, aspartate aminotransferase, and sorbitol dehydrogenase activities and bile acid concentrations. Most cattle with biochemical evidence of liver disease and severe HL had been lactating for 14 or more days. Cows that required intensive treatment inconsistently had serum biochemical evidence of liver disease. Although cattle with severe HL had significantly higher serum bilirubin concentrations and aspartate aminotransferase and sorbitol dehydrogenase activities than cattle with less severe lipidosis, the specificity of abnormally high serum sorbitol dehydrogenase activity or bilirubin concentration for severe lipidosis was only 8%. Abnormally high serum aspartate aminotransferase activity was 83% sensitive and 62% specific for severe lipidosis. Serum glucose and total carbon dioxide concentrations were significantly lower in cattle with severe lipidosis than in those with mild or moderate lipidosis, and low serum glucose or total carbon dioxide concentrations were rare in cattle without severe lipidosis. From these data, we conclude that the use of a single biochemical or histopathologic criterion to define severity of disease or degree of liver compromise in anorectic, ketotic cows results in the misidentification of many animals.