The aim of this investigation was to study secular trends in long-term survival following myocardial infarction (MI). Five thousand one hundred and fifty-seven consecutive cases of MI in 3942 patients were recorded in a well-defined region in the study period 1977-1988. The study period ended before thrombolytic therapy was introduced in the hospital. One and 5-year survival (+/- 95% confidence limits) was 61 +/- 2% and 42 +/- 2% in 1977-1980. These figures changed to 61 +/- 2% and 44 +/- 2% in 1981-1984, and to 64 +/- 2 and 46 +/- 2% in 1985-1988. The improvement with time was statistically significant (P < 0.001). In a Cox proportional hazard model, time of infarction was an independent predictor of survival. Patients were subdivided into a high risk group suffering from either congestive heart failure or cardiac arrest during hospitalization, and a low risk group without these complications. Year of infarction was without importance in the high risk group but highly significant in the low risk group. Long-term survival following MI gradually improved prior to the introduction of thrombolytic therapy. The improvement was confined to low risk patients without cardiac arrest or congestive heart failure.