Background There is accumulating evidence that thrombolytic therapy is underused among eligible patients with acute myocardial infarction. We sought to determine whether potential errors in electrocardiographic diagnosis might be a contributing factor. Methods Seventy-five electrocardiograms were interpreted on 2 separate occasions by 3 cardiologists. Two criteria were compared for thrombolysis eligibility: (1) measurement of ≥1 mm ST-segment elevation in 2 contiguous leads (measured) and (2) criterion 1 plus the subjective opinion that the changes represented acute transmural injury (interpretive). The results were compared with computerized interpretations by the Marquette 12SL system. Results Raw agreement and agreement corrected for chance between raters for both criteria were excellent and tended to be better for interpretive compared with measured criteria (κ = 0.89 vs 0.78, respectively). Strict reliance on measured electrocardiographic criteria alone would have resulted in overuse of thrombolysis among all 3 raters. Based on the consensus opinion, the absolute overuse of thrombolysis would have been approximately 15% (P <.0034). The computer algorithm had a specificity of 100% and a sensitivity of 61.5%. Reliance on the computerized interpretation alone would have lead to underuse of thrombolytic therapy compared with consensus opinion (21.3% vs 34.6%; P <.005). Conclusion Agreement for suspected acute myocardial infarction tended to be better when the appearance of the ST segments was added to measurable ST elevation criteria. Strict reliance on measurable criteria may lead to the inappropriate overuse of thrombolysis. Although the Marquette 12SL system has excellent specificity, it has poor sensitivity for the diagnosis of thrombolysis-eligible AMI. Reliance on computerized electrocardiographic interpretation would lead to the inappropriate underuse of thrombolytic therapy in situations in which qualifying electrocardiographic criteria are actually met. (Am Heart J 2000;140:221-6.)