The specificity of a system measuring cell-mediated cytotoxicity as effector-induced target cell detachment from plastic recently adopted to study autologous hepatocyte killing in liver disease, was examined in 17 HBsAg positive liver patients whose hepatocytes (after biopsy digestion with collagenase) were incubated in Terasaki plates with the corresponding blood lymphocytes over two days. The hepatocyte viability and the specificity of the effectors were evaluated as determinants of the clinical value of the test. We found that: (a) hepatocytes in all experiments showed membrane damage owing to the lytic action of collagenase on the small liver core; (b) patients' lymphocytes detached diseased autologous hepatocytes more efficiently than did normal lymphocytes with healthy hepatocytes; (c) in eight patients cytotoxicity appeared equally distributed between a population enriched in T cells and one enriched in non-T cells; yet the mean cytotoxic index of the latter subset was higher than that of the former; (d) cytotoxicity was not blocked by the addition of either aggregated IgG or purified HBsAg; (e) protein synthesis seemed required to promote hepatocyte detachment, for lymphocytes treated with Actinomycin D were no longer active. Poor target viability detracts from the specificity and the clinical value of the test, that therefore turns out to be a major problem of liver cell culture.