Clinical and biochemical data obtained from 163 patients who had taken an overdose of paracetamol were examined to determine which factors or measurements were of value in predicting the severity of ensuing liver damage early after ingestion of tablets. Although the overall severity of hepatic necrosis was found to increase with the dose of paracetamol ingested, correlation was not sufficiently close to provide an accurate prognostic index in individuals. Severe hepatic damage was less likely if the patient had vomited or had a stomach wash-out within 6 hr of overdose. The plasma concentrations of paracetamol, measured at known times after overdose, distinguished those who developed hepatic dysfunction from those who did not, but there was a poor correlation, particularly in the first 6 hr after ingestion of tablets, between these values and the severity of ensuing liver damage. Estimates of early plasma paracetamol half-lives from three or more samples taken within 4 hr of admission showed that all patients developing moderate or severe liver damage had half-lives greater than 4 hr, but this was also the case in nearly one-third of those with minimal liver lesions only. It is concluded that there is no completely reliable early prognostic test for individual patients with paracetamol overdose. If each patient is selected for treatment with cysteamine (mercaptamine) or other agents on the basis of plasma paracetamol levels, up to 30% may receive this agent who are at risk from trivial hepatic damage only.