The authors conducted a mailed questionnaire survey of a 5% sample of the cohort of 20,643 people officially recognized by the Spanish government as having had toxic oil syndrome, a previously undescribed illness that was epidemic in Spain in 1981. After three mailings of a letter and questionnaire, responses for only 66% of the sample had been received. Nevertheless, responses were obtained from virtually all remaining patients (or surrogates for them in the cases of patients that had died) when they were sought by telephone. In 1981, there was clear-cut excess mortality in the cohort (standardized mortality ratio [SMR] 6.51; 95% confidence interval [CI]: 3.92-10.17). During the period January 1982 through 7 March 1988, there was no statistically significant overall mortality excess except during the period 1982-1983 among people aged < 65 years (SMR 2.26; 95% CI: 1.03-4.29). Toxic oil syndrome substantially altered the patterns of mortality among affected people. Analysis of deaths by cause among the TOS cohort will be useful for further evaluation of the long-term impact of the TOS epidemic.