Two hundred and twenty patients with clinical stage I, II, or III carcinoma of the breast who presented for mastectomy were prospectively evaluated for hepatic metastases with abdominal sonography and routinely available liver enzymes (LE). Both abdominal ultrasound (US) and LE were assessed. There were two false-positive sonogram, both with normal enzymes, in patients with Stage I and II. There was one positive sonogram at presentation in a patient with Stage III, confirmed by needle biopsy, with a lactate dehydrogenase (LDH) elevation. The metastatic yield was low with ultrasound, but other findings were revealed. LE were not useful in the preoperative diagnosis of hepatic metastases, demonstrating a low specificity. Thirty-three patients demonstrated abnormalities of Le. Twenty-seven of 33 had elevation of a single enzyme, LDH being the most common. Three patients had an elevation of four enzymes preoperatively. Hepatic metastases were not diagnosed in these three patients preoperatively on ultrasound or biopsy. The patients were observed with physical examination (PE) and LE for a mean of 26 months. During the follow-up study, 0.9 per cent of the patients with Stage I, 3.4 per cent with Stage II and 12.5 per cent with Stage III had hepatic metastases develop. In patients with these metastases, involvement of the liver was suggested by elevated LE In 66 per cent, on PE in 42 per cent and by history in 28 per cent. LE were the first indication of liver involvement in two patients. We are in agreement with previous authors for abandonment of hepatic imaging in the preoperative assessment because of the low yield for metastases. Forty-eight thousand dollars would have have been spent on this cohort ot diagnose one instance of hepatic metastasis preoperatively. US may be efficacious in the subgroup of patients with grossly elevated LE or a PE suggestive of hepatic involvement. Those patients who demonstrate metastases by us should have histologic confirmation if treatment would thereby be altered.