It has been suggested that organic solvent exposure may contribute to an increased risk of blood disorders, neurological, liver and renal disease, and cancer. A meta-analysis has been performed of 55 published mortality studies which involved solvent exposure and provided standardised mortality ratios (SMR) or relative risk (RR). The combined results showed the overall SMR to be 86.7 (95% confidence interval [Cl] = 83.7-89.9), while that for all sites of cancer was 92.3 (Cl = 87.5-97.4). Risk of death from leukaemia was increased (SMR = 112.2, Cl = 101.6-146.9) as was that from cancer of liver and biliary passages (SMR = 119.7, Cl = 104.4-137.2), even though the risk of death from cirrhosis was reduced (SMR = 81.5, Cl = 68.1-97.4). No excess risk of death from other diseases has been found. The favourable mortality might be from a "healthy worker effect', but the increase in death from liver cancer in the absence of excess deaths from cirrhosis is biologically plausible and justifies further investigation. The increase in mortality from leukaemia is likely to have been associated with exposure to benzene.