Patients undergoing percutaneous coronary revascularization (PCR) for narrowed saphenous vein grafts (SVGs) have a high incidence of subsequent cardiac events, but the relative contribution of treated and untreated SVGs, and of native coronary narrowings to late events is uncertain. This study evaluated the role of progression of SVG disease at untreated sites to cardiac events in these patients. All patients with successful PCR of SVG lesions who were enrolled in clinical trials with mandated repeat angiography from 1990 to 1994 were studied. One hundred three patients (age 63 +/- 8 years, 82% men, ejection fraction 54 +/- 12%, graft age 8 +/- 4 years), contributing 1,095 analyzable 15- to 25-mm SVG segments were followed 29 +/- 13 months (4 patients were lost to follow-up). Actuarial event-free (death, myocardial infarction, bypass surgery, or PCR) and overall survival at 12 months were 47 +/- 5% and 94 +/- 2%, respectively. Fifty-six percent of all early (< or = 12 months) events resulted from ischemia from recurrence at initially treated SVG sites, 26% at nontreated SVG sites, and 14% at nontreated native coronary sites. By 36 months, event-free and overall survival were 25 +/- 6% and 86 +/- 4%, respectively. Events occurring > 12 months after initial treatment resulted most frequently from ischemia from progression of narrowing at untreated SVG sites (46%). Ischemic events from initially untreated SVG sites were correlated with initial percent stenosis (initial, 41% to 50%; 45% events, 31% to 40%; 18% events, < or = 30%; 2% events, p <0.001) and reference SVG diameter (p = 0.003). Recurrent ischemic events from initially treated SVG sites were independently correlated with initial percent stenosis (initial > 75%; 43% events, 50% to 75%; 27% events, < 50%; 18% events, p = 0.01), but not with final percent stenosis. The frequent occurrence of events from nontreated 41% to 50% stenoses suggests a need for increased surveillance in patients with these lesions. The low incidence of events from initially treated lesions < 50% suggests that the hypothesis that "nonsignificant" 41% to 50% lesions might best be treated at the time other more severe narrowings are treated should be examined.