We compared central laboratory with local determinations of key clinical measurements in three VA Cooperative Studies. Electrocardiographic evidence of new myocardial infarction was assessed in the study of Aspirin Therapy and Unstable Angina, ejection fraction measurement in the Coronary Artery Bypass Surgery Trial and lesion size in the Angioplasty Compared with Medicine (ACME) Trial. The findings in the Aspirin Trial indicated that central coding of all serial electrocardiograms in 1266 patients to detect new acute myocardial infarction by computer algorithm was not cost-effective when compared with the local investigator's diagnosis on the basis of a central Electrocardiographic Committee as the reference standard. In the other two trials, the contribution of the central laboratories was important because the assessments of the local investigators generally underestimated the degree of abnormality in the Bypass Trial and overestimated it in the ACME Trial. The VA results have clearly demonstrated two cases in which the decision for central evaluation was prudent, but one case in which it was not cost-effective. These equivocal findings indicate the need to evaluate the contribution of central laboratories when used as an adjunct to local determinations. Such evaluations may provide guidelines for decision-making in the design of future trials.