OBJECTIVES: To extend the follow up of a cohort of 14,457 aircraft maintenance workers to the end of 1990 to evaluate cancer risks from potential exposure to trichloroethylene and other chemicals. METHODS: The cohort comprised civilians employed for at least one year between 1952 and 1956, of whom 5727 had died by 31 December 1990. Analyses compared the mortality of the cohort with the general population of Utah and the mortality and cancer incidence of exposed workers with those unexposed to chemicals, while adjusting for age, sex and calendar time. RESULTS: In the combined follow up period (1952-90), mortality from all causes and all cancer was close to expected (standardised mortality ratios (SMRs) 97 and 96, respectively). Significant excesses occurred for ischaemic heart disease (SMR 108), asthma (SMR 160), and cancer of the bone (SMR 227), whereas significant deficits occurred for cerebrovascular disease (SMR 88), accidents (SMR 70), and cancer of the central nervous system (SMR 64). Workers exposed to trichloroethylene showed non-significant excesses for non-Hodgkin's lymphoma (relative risk (RR) 2.0), and cancers of the oesophagus (RR 5.6), colon (RR 1.4), primary liver (RR 1.7), breast (RR 1.8), cervix (RR 1.8), kidney (RR 1.6), and bone (RR 2.1). None of these cancers showed an exposure- response gradient and RRs among workers exposed to other chemicals but not trichloroethylene often had RRs as large as workers exposed to trichloroethylene. Workers exposed to solvents other than trichloroethylene had slightly increased mortality from asthma, non- Hodgkin's lymphoma, multiple myeloma, and breast cancer. CONCLUSION: These findings do not strongly support a causal link with trichloroethylene because the associations were not significant, not clearly dose-related, and inconsistent between men and women. Because findings from experimental investigations and other epidemiological studies on solvents other than trichloroethylene provide some biological plausibility, the suggested links between these chemicals and non-Hodgkin's lymphoma, multiple myeloma, and breast cancer found here deserve further attention. Although this extended follow up cannot rule out a connection between exposures to solvents and some diseases, it seems clear that these workers have not experienced a major increase in cancer mortality or cancer incidence.