Drug susceptibility testing is usually performed when Mycobacterium tuberculosis organisms are recovered from Asian immigrants or from patients whose sputum remains culture positive despite several months of antituberculosis medication. Alcoholism, previous antituberculosis treatment, history of adverse reactions to previous treatment, and patient unreliability have also been suggested as risk factors, but the ability of these factors to predict the presence of drug-resistant organisms has not been assessed. Starting in January 1980, the Washington State Tuberculosis Laboratory began testing every positive M. tuberculosis culture for drug resistance. This enabled us to prospectively evaluate the sensitivity and specificity of 7 risk factors for drug resistance in consecutive patients. From January 1, 1980, through December 31, 1982, cultures from 803 patients were positive for M. tuberculosis; 766 of these (95%) were tested for drug resistance. The incidence of resistance was 13% (101/766). Of the 101 patients with drug-resistant organisms, 61 (60%) were Asian, supporting the need to routinely test Asian immigrants for drug-resistant disease. The ability of the other risk factors to separate the 40 non-Asians (40%) with drug-resistant disease from the group of 513 non-Asians with drug-sensitive organisms was poor. All risk factors were insensitive and had a high false positive rate. These results demonstrate that the presence of drug-resistant disease in non-Asians cannot accurately be predicted. This finding suggests that all cultures of M. tuberculosis should be tested for drug sensitivity, and that in areas where the incidence of drug resistance is sufficiently high, initial treatment should be with 3 drugs until drug susceptibility is known.