Quantitative analysis of ST-segment deviations (STdev) and their resolution by treatment (STR; calculated from a combined sum of STdev in multiple leads) are used in trials on reperfusion for myocardial infarction (MI). Unreadable or unavailable electrocardiogram (ECG) leads are a common reason for exclusion, decreasing the statistical power of the trials. We developed mathematical formulas for reconstruction of immeasurable STdev based on STdev from other available leads on the 12-lead ECG. Formulas were deducted from a database of computer-assisted STdev measurements in 2 ECGs (baseline and 180 min after thrombolysis) of 1121 pts. Their accuracy was later evaluated on a second dataset of 377 pts. Acceptable fits could be derived for absent single leads, or for groups of absent limb leads (I-II-III or aVL-aVF). The intraclass correlation coefficient between real and calculated STdev was >or= 0.80 for each (0.77 for V1 in inferior MI). The correlations between STR calculated from original data and from reconstructed STdev were very strong (all intraclass correlation >or=0.97), and discordance in STR subgroup categorization occurred in <or=10% of pts in all but one of the scenarios (I-II-III substituted in 180 min ECG in inferior MI). Scenarios with multiple missing precordial leads however are not substitutable, nor are calculated STdev reliable for STR evaluation in only the lead with highest elevation in baseline. STdev reconstruction formulas can reliably be used in trials where analysis of aggregate STR is an endpoint. Reliable substitution can significantly increase the number of evaluable patients and therefore strengthen the statistical power of these trials.