In 198 patients with acute myocardial infarction serial measurements of plasma creatine kinase isoenzyme MB (CK MB) were performed at four hour intervals. In every patient, maximal CK MB activity (peak activity) was compared with calculated total release per litre plasma. In 28 patients (group 1) sufficient plasma samples were available for calculation of the apparent first order inactivation constant kd. Mean apparent kd in group 1 patients was 0 . 085 +/- 0 . 018 h-1 (mean +/- SD). Total release in group 1 was calculated with individual apparent kd values (Q) and with the mean kd value (Q*). In the remaining 170 patients (group 2), Q* only was calculated. A linear relation between peak activity P and total release (both Q and Q*) was found, extending over the whole range of CK MB peak activities that are routinely observed (4-216 U/l). It was immaterial whether a one or a two compartment model was used: both yielded a close linear relation. Though the mean ratio between Q* and peak activity depends on the value of kd chosen for calculation of total release (the ratio increasing with increasing kd), linearity between peak activity and Q* was found for any value of kd up to 0 . 4 h-1. In group 1, shapes of calculated CK MB release curves Q*(t), expressed relative to maximal release Q(40), were sufficiently similar so as to be superimposable; the section of the release curves extending from 12 hours before until two hours after peak time could be tentatively described by a linear time course with a slope of 4 . 2 +/- 0 . 5% per hour (mean +/- SD). We conclude that peak activity of CK MB is a reliable estimate of cumulative CK MB release and may be clinically more practicable than calculation of Q(40). Both the similarity and the large apparently linear section of the calculated enzyme release curves possibly permit early prediction of Q(40), with acceptable precision.