An American Type Culture Collection reference strain and eight clinical strains of Mycobacterium tuberculosis, all of which were susceptible to isoniazid (INH) (mean MIC, 0.06 mg/liter) and negative for the Ser315Thr katG mutation, were left in their BACTEC 12B vials (for use with the BACTEC 460-TB method) containing 0.1 mg of INH per liter for periods of up to 28 days after the completion of the antibiotic susceptibility test. Each eventually grew to levels compatible with those of INH-resistant strains. Successive passages in INH-containing BACTEC 12B vials and onto solid media showed that the resistance noted above was maintained. Successive passages of these M. tuberculosis strains in which INH resistance had been induced into BACTEC 12B vials or solid media containing stepwise increases in INH concentrations eventually yielded organisms resistant to 20 mg of INH per liter. Transfer of cells in which INH resistance had been induced to drug-free medium followed by repeated passages in that medium eventually yielded organisms whose susceptibility to INH was identical to that of the original parent strains. The cycle of induced INH resistance could be repeated with these now INH-susceptible cells. The use of M. tuberculosis identification probes and IS6110-based restriction fragment length polymorphism analyses of cultures throughout the induction of INH resistance and the reversal of resistance in drug-free medium eliminated the possibility that the culture was contaminated or that the initial specimen had a mixed type of infection. Induced high-level resistance to INH (20 mg/liter) could be reduced 100-fold with a subinhibitory concentration of reserpine but not with verapamil. These results collectively suggest that high-level resistance to INH can be induced in INH-susceptible M. tuberculosis strains by the induction of a reserpine-sensitive efflux mechanism.