BACKGROUND Cardiac allograft arteriopathy often limits long-term survival in transplantation recipients but has been difficult to detect by standard diagnostic methods. Because of the diffuse nature of transplantation coronary disease, we postulated that a lung/heart ratio during dipyridamole thallium imaging might better predict arteriopathy-related complications than diagnostic methods that detect discrete luminal stenoses. METHODS AND RESULTS Sixty-six unselected heart transplantation recipients were evaluated with annual coronary arteriograms, endomyocardial biopsy, and intravenous dipyridamole thallium testing (initial study group). The mean lung/heart ratio on an anterior planar image was 0.40 for all patients; therefore <0.40 was arbitrarily defined as normal. After October 1992, 98 patients were tested (validation study group) and a lung/heart ratio cutoff of 0.40 was evaluated prospectively. Coronary end points were defined as (1) at least 1 coronary artery stenosis >/=50% of the luminal diameter, (2) sudden cardiac death, and (3) acute myocardial infarction. Stepwise logistic regression analysis was performed to identify independent predictors of future coronary end points. For the initial study group, the lung/heart ratio on the first annual thallium study was the only independent predictor of subsequent cardiac end points (0.47 +/- 0.13 [SD] with end points vs 0.38 +/- 0.11 without end points, P <.05). For the validation study group, independent predictors of subsequent coronary events included the lung/heart ratio and the radionuclide left ventricular ejection fraction. No patient with a lung/heart ratio <0.40 and a left ventricular ejection fraction >/=0.50 developed a cardiac event during 21 +/- 11 months of follow-up. CONCLUSIONS A lung/heart ratio >/=0.40 on dipyridamole thallium testing is a sensitive predictor of coronary events after heart transplantation. Patients with heart transplantion who have a lung/heart ratio <0.40 and normal systolic left ventricular function are at low risk for subsequent coronary events and may not require annual surveillance by coronary arteriography.