BACKGROUND There is poor adherence with tuberculosis preventive therapy among patients released from short-term correctional facilities, leading to recommendations against screening for latent tuberculosis infection (LTBI) in this setting. OBJECTIVE To assess adherence to isoniazid preventive therapy (IPT) following release from short-term correctional facilities, and to estimate the cost-effectiveness of this practice. METHODS Records of individuals referred for IPT from the Connecticut Department of Corrections to the City of Hartford Chest Clinic between January 1993 and June 1997 were reviewed. The data abstracted included demographics, adherence to IPT, and the duration of IPT completed before release from prison. An analysis was performed to determine the cost-effectiveness of this program. RESULTS A total of 168 records were reviewed. The mean duration of IPT completed before release from prison was 8 weeks. Eighty-six subjects (57%) never came to clinic after release. Of the 64 subjects (43%) who attended clinic at least once, 35 subjects (55%) completed IPT and 29 subjects (45%) were unavailable for follow-up before completing therapy. Thirty-three of the 64 subjects (52%) who attended the clinic had to be restarted on IPT due to a prolonged lapse in therapy prior to the first visit. We estimate that $32,866 was spent on this program, but $42,093 in future costs associated with reactivation tuberculosis was prevented. CONCLUSIONS Adherence with IPT is poor in patients released from short-term correctional facilities. Nonetheless, this program was cost-effective. An alternative strategy may be to screen for LTBI among inmates of short-term correctional facilities but withhold IPT in inmates expected to be released before therapy would be completed. Instead, these inmates could be referred to an appropriate clinic after release. Prophylaxis may be started in subjects who keep an initial clinic appointment after release.