An improved radio-immunoassay using an antiserum directed towards the N-terminal part of the endogenous opioid peptide beta-endorphin 1-31 (beta-EP) was validated and applied to a study of beta-EP in plasma during ischaemic pain. Experimental ischaemic pain induced in seven healthy volunteers by the submaximal effort tourniquet test did not change plasma beta-EP or adrenocorticotrophin. Plasma beta-EP was determined in 21 patients with acute myocardial infarction (AMI) and in seven patients with unstable angina pectoris. Plasma beta-EP was 4.9 fmol/ml with 95% confidence limits, 3.2-7.8 fmol/ml in AMI patients at admittance, and 2.9 (2.0-3.4) fmol/ml one week later in stable and pain-free condition (p less than 0.05). The level in 49 healthy persons was 2.8 (2.4-2.9) fmol/ml. Elevated beta-EP levels were found in five AMI patients with cardiogenic shock and in four AMI patients dying within 24 h after admittance compared to the rest of AMI patients (p less than 0.02). beta-EP was not elevated during unstable angina pectoris, although pain scores were similar to AMI. The AMI group revealed a significant, although weak, positive correlation between plasma beta-EP and pain score (Spearman r = 0.49, p less than 0.05), while there was no correlation during unstable angina pectoris. beta-EP was not correlated to the amount of morphine required within the 48 h after admittance of AMI patients. We conclude that the increase of beta-EP in plasma during AMI may be due to stressful factors other than ischaemic pain and that it is questionable whether beta-EP in plasma is related to antinociception.