The ability of computed tomography (CT) to detect mediastinal lymph node metastases from nonsmall cell bronchogenic lung cancer is highly controversial, as evidenced by reported accuracies ranging from 0.35 to 0.95 over the past eight years. We examined all studies on this matter published between January 1980 and April 1988, both to describe the overall experience and to identify characteristics (study design and methodology and CT scan techniques) that influenced reported accuracy. Of 79 relevant publications, 37 were excluded because they were review reports, assessed small cell lung cancer, or contained insufficient evidence to construct a contingency table (CT result versus node histology). The pooled, unweighted (weighted) results based on the remaining 42 studies were as follows: sensitivity, 0.79 (0.83); specificity, 0.78 (0.81); accuracy, 0.79 (0.81). Using a node size greater than 1.0 cm to define a "positive" CT result, as compared to a smaller diameter, was associated with significantly higher specificity, 0.89 versus 0.76, and accuracy, 0.86 versus 0.75 (p less than or equal to 0.005), but not sensitivity, 0.79 versus 0.75. The observed differences in accuracy between a fourth generation CT (0.83) and either a third or a second generation CT, (0.77 and 0.78, respectively) were not significant at p less than 0.05. No characteristics, either singly or in combination, resulted in accuracies exceeding 0.86. There exists random variation of individual study results around an overall mean accuracy of only 0.79, which is marginally improved by advances in CT technology and methods. Significant advances in the noninvasive detection of lymph node metastases must await an approach fundamentally different from CT-determined node size.