This study was designed to determine whether an integrated screening program could be implemented that identified persons with latent tuberculosis infection among jail inmates and produced a high rate of completion of isoniazid preventive therapy (IPT) in those persons after their discharge from short-term detention, by means of community-based directly observed preventive therapy (DOPT). From June 1, 1992 through December 31, 1994 inmates in the King County Jail who were from populations at high risk of tuberculosis were screened by means of the tuberculin skin test and those with latent tuberculosis infection were offered IPT. Among 262 inmates receiving IPT upon release from jail, 105 (40%) could not be located after release. Among another 105 enrolled on DOPT, 63 (60%) completed therapy. Among 52 who chose self-supervised isoniazid therapy after release, 15 (29%) completed therapy (chi-square = 13.50, p = 0.0002). Among persons with latent tuberculosis infection detected during screening at a county jail, a postrelease DOPT program resulted in a high rate of immediate loss to follow-up and a low rate of completion of therapy. Based on these results, we suggest that funds for TB control, if limited, should not be diverted to jail-based screening and postrelease DOPT.