Investigation of people exposed to cases of infectious tuberculosis (contact investigation) is key to tuberculosis control in countries with low tuberculosis incidence. However, in countries in which the incidence of tuberculosis is high, contact investigation is not commonly done. Increasing concerns about the failure to meet case-detection targets and about the spread of drug-resistant Mycobacterium tuberculosis have prompted a reassessment of the potential benefits of contact investigation. We did a systematic review to determine the yield of household contact investigation. The yield for all tuberculosis (bacteriologically confirmed and clinically diagnosed) was 4.5% (95% CI 4.3-4.8, I(2)=95.5%) of contacts investigated; for cases with bacteriological confirmation the yield was 2.3% (95% CI 2.1-2.5, I(2)=96.6%). Latent tuberculosis infection was found in 51.4% (95% CI 50.6-52.2, I(2)=99.4%) of contacts investigated. The substantial heterogeneity in all analyses indicated high variability among studies that was not accounted for by subgroup analyses. These results suggest that contact investigation merits serious consideration as a means to improve early case detection and decrease transmission of M tuberculosis in high-incidence areas.