In studies assessing the trends in coronary events, such as the World Health Organization (WHO) MONICA Project (multinational MONItoring of trends and determinants of CArdiovascular disease), the main emphasis has been on coronary deaths and non-fatal definite myocardial infarctions (MI). It is, however, possible that the proportion of milder MIs may be increasing because of improvements in treatment and reductions in levels of risk factors. We used the MI register data of the WHO MONICA Project to investigate several definitions for mild non-fatal MIs that would be applicable in various settings and could be used to assess trends in milder coronary events. Of 38 populations participating in the WHO MONICA MI register study, more than half registered a sufficiently wide spectrum of events that it was possible to identify subsets of milder cases. The event rates and case-fatality rates of MI are clearly dependent on the spectrum of non-fatal MIs, which are included. On clinical grounds we propose that the original MONICA category "non-fatal possible MI" could be divided into two groups: "non-fatal probable MI" and "prolonged chest pain." Non-fatal probable MIs are cases, which in addition to "typical symptoms" have electrocardiogram (ECG) or enzyme changes suggesting cardiac ischemia, but not severe enough to fulfil the criteria for non-fatal definite MI. In more than half of the MONICA Collaborating Centers, the registration of MI covers these milder events reasonably well. Proportions of non-fatal probable MIs vary less between populations than do proportions of non-fatal possible MIs. Also rates of non-fatal probable MI are somewhat more highly correlated with rates of fatal events and non-fatal definite MI. These findings support the validity of the category of non-fatal probable MI. In each center the increase in event rates and the decrease in case-fatality due to the inclusion of non-fatal probable MI was larger for women than men. For the WHO MONICA Project and other epidemiological studies the proposed category of non-fatal probable MIs can be used for assessing trends in rates of milder MI.