There is still debate over the relative merits of cytology and histology in diagnosing hepatocellular carcinoma in cirrhotic livers. Previous comparisons of the diagnostic accuracies of these two methods may have been biased by sampling errors due to multiple punctures. We compared the diagnostic accuracies of cytology and microhistology using tissue and cells from the same point in liver nodules subsequently proved to be hepatocellular carcinoma. A single ultrasound-guided liver-nodule biopsy was obtained with a 20- to 21-G cutting needle from 131 cirrhotic patients. The solid portion of samples was used for microhistology; the remainder was subjected to smear cytology. The results of each type of examination were expressed as true positive, nonspecific malignancy, false negative, or inadequate for diagnosis. No false-positive diagnoses were made in 13 benign lesions. In 118 HCC nodules (particularly those <30 mm in diameter), cytology provided a significantly higher percentage of correct diagnoses (85.6%) that was only slightly inferior to that based on results of both studies (89.8%). The single-biopsy technique generally provides adequate tissue for histology and cytology specimens with a high cellularity. It reduces both the cost and the risks of fine-needle biopsy diagnosis of hepatocellular carcinoma.