PURPOSE: To review the accuracy of the history and physical examination, clinical assessment, and noninvasive diagnostic tests for the diagnosis of deep vein thrombosis (DVT). BACKGROUND: Each year, over 250,000 patients are hospitalized in the United States for the treatment of acute DVT, and many more are diagnosed while hospitalized for surgical procedures or medical illness or while residing in long‐term care facilities. Prompt diagnosis and treatment are essential to prevent the short‐term sequelae of pulmonary embolism and death and the long‐term complications of recurrent venous thromboembolism and the postthrombotic syndrome. Knowledge of the accuracy of elements of the history and physical examination helps clinicians assess the probability of DVT before ordering confirmatory diagnostic studies or instituting therapy. Several studies have found that the incidence of venous thrombosis increases with age. The important contribution of age was demonstrated in a study of Swedish men that found the cumulative probability for venous thromboembolism to be relatively low (0.5%) by the age of 50 but substantial (10.7%) by the age of 80. The case‐fatality rate also increases with age; mortality increases exponentially beginning at the age of 40. Given the advancing age of the population of the United States, clinicians can expect to be assessing patients, and particularly older patients, for suspected DVT with increasing frequency. The authors have reviewed the literature to describe the accuracy of various elements of the history and physical examination, the combination of these findings, and noninvasive tests in the diagnosis of DVT. DATA SOURCES: A structured Medline search was performed (limited to English‐language studies from 1966 to 1997) to identify studies examining the clinical assessment of patients with suspected DVT. Bibliographies of the retrieved articles were reviewed for additional relevant studies. STUDY SELECTION CRITERIA: Studies examining the accuracy of the physical examination, clinical assessment, or diagnostic tests for the diagnosis of DVT were selected for complete review. DATA EXTRACTION: Of 115 articles retrieved, 68 were considered relevant for analysis and were completely reviewed. The selected studies were graded based on the quality of their methods using a scoring system described by Sackett et al. Data for individual trials are presented separately. Meta‐analyses were not performed. The prevalences of symptoms and signs of DVT are reported for patients with and without DVT. Data are presented as odds ratios (ORs) for risk factors and as sensitivity, specificity, and likelihood ratios (LRs) for the clinical assessment of DVT. The derivation and validation of a clinical decision rule is described in detail. The results for the clinical decision rule are reported as positive predictive values and LRs. ORs, LRs, sensitivities, and specificities are reported with 95% confidence intervals (CIs).