The incidence of tuberculosis in the Native population in Alaska (92.2/100,000) continues to be higher than the U.S. tuberculosis incidence (9.3/100,000). This paper describes tuberculosis surveillance in Alaska and estimates the "predictive value positive" for cases recorded by the surveillance system during 1986. (Predictive value positive refers to the proportion of all people identified who actually have the disease.) Active surveillance components included an itinerant chest clinic and survey chest roentgenography program, epidemiologic case investigations, and skin testing. Passive components included cases reported by health-care providers and the U.S. Immigration and Naturalization Service. Half of all new cases were reported by health-care providers. Only 42% of cases were bacteriologically confirmed (vs 87% nationwide) suggesting that some cases may have been false positives. many cases did not have a contact investigation completed. Recommendations to improve the surveillance system include: 1) following the Centers for Disease Control/American Thoracic Society diagnostic criteria more closely, 2) increasing the number of cases which have contact investigations completed, and 3) targeting surveillance activities towards high risk subgroups such as Southeast Asian immigrants.