It is maintained that pulmonary angiography is required for confirmation or exclusion of pulmonary embolism in the majority of patients suspected of having pulmonary embolism. The aim of this study was to reappraise the role of perfusion scan in conjunction with clinical assessment in the diagnosis of pulmonary embolism and to identify subsets of patients in whom angiography is strictly required for definitive diagnosis. At the time of referral, each of 252 consecutive patients was assigned a clinical probability of pulmonary embolism (very likely, possible, unlikely). Perfusion scan was subsequently obtained and assigned to one of the following categories: (1) normal; (2) near normal; (3) single or multiple wedge-shaped perfusion defects compatible with pulmonary embolism (PE+); (4) perfusion defects other than wedge-shaped not compatible with pulmonary embolism (PE-). By protocol, angiography had to be obtained in all patients with abnormal scan (PE + and PE-). The protocol was completed in 176 patients (107 with normal/near normal and 69 with abnormal scan in whom a definitive diagnosis was reached). The overall rate of correct clinical classification was 86%. Sensitivity and specificity of PE+ perfusion scan were 89 and 92%, respectively. Pulmonary embolism was present in all 37 patients with very likely or possible clinical presentation and PE+ scan (positive predictive value 100%) and in 2 of 17 cases with low likelihood of pulmonary embolism and PE- scan (negative predictive value of 88%). These preliminary results indicate that pulmonary embolism can be diagnosed noninvasively in the majority of cases and that angiography is strictly required only for a minority of patients (21% in this study) in whom clinical and perfusion scan assessment are discordant.