In a first experiment, the efficacy of a 6-month course of isoniazid (INH) alone in comparison with 2-month courses of rifampin (RMP) alone, RMP + pyrazinamide (PZA), or RMP + PZA + INH as preventive therapy of tuberculosis was evaluated in the mouse. To simulate the infected, non-diseased state of humans, a nonreplicating bacillary population of limited size was developed in the mouse. Mice were vaccinated intravenously with 2.74 log10 M. bovis BCG and infected a month later with 3.38 log10 M. tuberculosis. Treatment began 2 wk after infection when the mean size of the M. tuberculosis population was 4.98 +/- 0.26 log10 cfu in the spleen. After 2 months of therapy, the proportion of mice with positive spleen cultures was 100%, 50%, 0%, and 20% in those animals treated, respectively, with INH alone, RMP alone, RMP + PZA, or RMP + PZA + INH. After 6 months of therapy with INH alone, the proportion of mice with positive spleen cultures was 38%. In order to confirm the extreme activity of the combination RMP + PZA and to assess the value of 3 months of therapy with RMP, a second experiment was performed following similar procedures. On completion of treatment, the proportion of mice with positive spleen cultures was 100%, 20%, 0%, and 80% in those animals treated, respectively, for 6 months with INH alone, 3 months with RMP alone, or 2 months with RMP + PZA, or RMP + PZA + INH.(ABSTRACT TRUNCATED AT 250 WORDS)