The objective of this study was to assess the performance of the clinical evaluation in detecting extrathoracic metastases compared with brain and abdomen CT and radionuclide bone scans in patients with newly diagnosed bronchogenic carcinoma. The included studies were selected using the MEDLARS database from 1977 through August 1992 as well as reference lists from published articles or abstracts. Studies eligible for consideration met six criteria. The most important criterion was that results of a clinical evaluation and a CT scan of the head or abdomen or a radionuclide bone scan, obtained during the initial evaluation of a patient with primary lung cancer, must be included. Data were categorized by the type of clinical evaluation performed and whether patients had a clinical evaluation suggesting metastases (positive) or not (negative). The negative predictive value (NPV) of the clinical evaluation was calculated in all studies. The sensitivity, specificity, and the positive predictive value (PPV) were calculated in studies including positive and negative clinical evaluation patients. Twenty-five studies are included in this analysis. A total of 3,089 imaging scans were obtained in the study patients after a clinical evaluation was performed. The mean NPV of the clinical evaluation for CT of the brain, abdomen, and radionuclide bone scan is 95, 94, and 89%, respectively. When an expanded clinical evaluation was performed, the NPV was even higher. The NPV was influenced by the prevalence of metastases, but still performed well in series with high prevalence rates.(ABSTRACT TRUNCATED AT 250 WORDS)