Ambient temperatures trigger hospitalisation, mortality, and emergency department visits for myocardial infarction (MI). However, nonoptimum temperature-related risks of fatal and nonfatal MI have not yet been compared. From 2007 to 2019, 416,894 MI events (233,071 fatal and 183,823 nonfatal) were identified in Beijing, China. A time-series analysis with a distributed-lag nonlinear model was used to compare the relative and population-attributable risks of fatal and nonfatal MI associated with nonoptimum temperatures. RESULTS: The reference was the optimum temperature of 24.3°C. For single-lag effects, cold (-5.2°C) and heat (29.6°C) effects had associations that persisted for more days for fatal MI than for nonfatal MI. For cumulative-lag effects over 0 to 21 days, cold effects were higher for fatal MI (relative risk [RR] 1.99, 95% confidence interval [CI] 1.68-2.35) than for nonfatal MI (RR 1.60, 95% CI 1.32-1.94) with a P value for difference in effect sizes of 0.048. In addition, heat effects were higher for fatal MI (RR 1.33, 95% CI 1.24-1.44) than for nonfatal MI (RR 0.99, 95% CI 0.91-1.08) with a P value for difference in effect sizes of 0.002. The attributable fraction of nonoptimum temperatures was higher for fatal MI (25.6%, 95% CI 19.7%-30.6%) than for nonfatal MI (19.1%, 95% CI 12.1%-25.0%). Fatal MI was more closely associated with nonoptimum temperatures than nonfatal MI, as evidenced by single-lag effects that have associations which persisted for more days, higher cumulative-lag effects, and higher attributable risks for fatal MI. Strategies are needed to mitigate the adverse effects of nonoptimum temperatures.