Clinical and experimental data have shown that after acute myocardial infarction there is a significant release of tumour necrosis factor alpha. Therefore, an attempt was made to correlate changes in serum tumour necrosis factor alpha concentrations with indices of infarct extent in patients with acute myocardial infarction. In 50 patients with acute myocardial infarction, blood samples for evaluation of tumour necrosis factor alpha and alpha-hydroxybutyrate-dehydrogenase were collected every 6 h until 120 h after admission. Infarct extent was estimated by clinical parameters such as the occurrence of heart failure and rhythm disturbances, by enzymatic methods such as cumulative release of alpha-hydroxybutyrate-dehydrogenase and imaging techniques, by late resting single photon emission tomography--201 thallium scintigraphy--using an extent score and by echocardiography using a wall motion index. The maximum change in serum tumour necrosis factor alpha after infarction (delta TNF) was calculated by subtracting tumour necrosis factor alpha concentration on admission from peak tumour necrosis factor alpha concentration. The average peak tumour necrosis factor alpha level was observed 84 h after admission (median: 12 pg.ml-1). Between the 72nd and the 96th h no significant changes in tumour necrosis factor alpha values were observed. Analysis of the data showed that larger delta (TNF) values were found to be associated significantly with signs of heart failure (P = 0.003), the presence of rhythm disturbances (P = 0.001), increased enzymatic infarct extent indicated by cumulative release of alpha-hydroxybutyrate-dehydrogenase (r = 0.74; P < 0.001), large myocardial perfusion defects measured with 201 thallium scintigraphy (r = 0.80; P < 0.001), and a considerable number of left ventricular wall motion abnormalities (r = 0.57; P < 0.001). In conclusion, delta (TNF) is a reliable method of assessing damage severity in the myocardium after acute myocardial infarction. As only two blood samples are necessary within 84 h, the method may be one of the more convenient for the assessment of infarct size in clinical practice.