D-dimer assay (DDA), measuring fibrin degradation products, was compared with lung scintigraphy (LS) in a prospective unselected series of 83 consecutive patients referred owing to suspicion of pulmonary embolism (PE). This patient series was also used to compare several methods of performing and interpreting LS images. The final diagnosis was established independently by a separate panel with all available information except for the result of DDA. D-dimer was determined by ELISA (threshold value 500 ng/ml). LS, including perfusion (.Q) and pseudo-ventilation (Technegas) (.V), was classified according to PIOPED, (1) immediately by the physician on duty, and (2) retrospectively by a blinded panel. A positive (19) or negative (61) diagnosis of PE was achieved in 80 patients, the prevalence of PE being 24%. Only one false-negative was noted on DDA (sensitivity=95%) but there were 42 false-positives (specificity=31%), resulting in a positive predictive value of 30% and a negative predictive value of 95%. Emergency and retrospective interpretations of LS were close (kappa=0.4). In a minority of patients, PE may be excluded with reasonable certainty if DDA is normal, resulting in a significant saving in terms of time and money.