To determine whether risk factors or angiographic features could be correlated with progression of coronary artery disease, we independently analyzed serial coronary angiograms in 65 symptomatic patients for progression without knowledge of their respective risk factors. An important design feature of this study was that observer variability of coronary lesion assessment was objectively evaluated and taken into account in defining progression of coronary artery disease. The reproducibility of lesion assessment varied with the severity of stenosis: moderate stenoses had the greatest intraobserver and interobserver variability. At initial study, 337 partial and 68 complete occlusions were identified. Of the 337 partial occlusions at risk of progression, 73 (22%) progressed; 13 (3.2%) of the initial lesions regressed. In the 65 patients with persistent or increasing angina, progression occurred in 51 (78%) over a mean period of 24 months. Apart from the tendency of high-grade stenosis( greater than or equal to 98%) to progress to complete occlusion, no measured clinical, laboratory, or angiographic variable showed any significant effect on progression to coronary artery disease, in this selected group of symptomatic patients these same risk factors did not predict its progression. We believe that (1) and definition of progression should incorporate the factor of observer variability and (2) therapeutic measures designed to influence progression of coronary artery disease should take into account this apparently unpredictable progression.