Of 84 cases with both radionuclide and computed tomographic studies of the liver, in 54 the results of the two examinations agreed, in 14 results disagreed, and in 16 cases one of the two tests gave equivocal results. Overall accuracy in detecting or ruling out focal liver disease was 79% for scintigraphy and 98% for computed tomography. In 13 of the 14 patients in whom studies did not agree, the scintigram was falsely negative. In 10 of these 13 false-negative scintigrams, small (less than 2 cm) solitary or multiple lesions were detected on computed tomography in patients who had normal liver function studies. In seven patients in whom both studies agreed, computed tomography provided additional, extrahepatic information that altered clinical management. Scintigraphy was 86% accurate in detecting liver disease when the patient had a gastrointestinal neoplasm, but only 74% accurate for nongastrointestinal neoplasm. Computed tomography was over 94% accurate in both situations. These data indicate an advantage of computed tomography as the initial screening examination for space-occupying lesions of the liver, especially in patients with a primary diagnosis of nongastrointestinal neoplasm.