Currently, the Category (CAT) II regimen is recommended for patients who have failed the CAT I regimen. We have determined before that prevalence of multidrug-resistant tuberculosis (MDR TB) is relatively high among these patients. On the other hand, the retreatment success rate with CAT II in CAT I treatment failures and defaults is nearly 50%. Therefore, we tried to find another strategy with a higher success rate. From January 2004 to November 2007, 105 patients with pulmonary TB, who failed a prior CAT I regimen or with more than one course of irregular anti-TB treatment, were included in this study, whereas five cases with nontuberculous mycobacteria were excluded. Drug susceptibility testing (DST), for first line anti-TB drugs, and polymerase chain reaction were performed. By the time of availability of DST that took 3 to 4 months, a pilot protocol consisted of isoniazid, rifampin, ethambutol, ofloxacin, cycloserine, and amikacin was started. Then therapeutic regimen was adjusted based on four categories of DST pattern: sensitive, non-MDR pattern, MDR pattern, and culture-negative. Sensitive patients received the standard CAT I regimen, non-MDR patients an individualized regimen based on DST, MDR patients a standard second-line regimen, and culture-negatives a standard CAT I plus a 6-month injectable agent. Treatment outcomes were categorized and analyzed. Forty-eight patients with prior CAT I treatment failure and 52 with more than one irregular treatment courses were included in the analysis. Six percent of subjects had confirmed HIV infection. Seventy-two percent of subjects were assigned to a good outcome and 28% were assigned to a poor outcome group. Seventeen percent were culture-negative. Regarding DST pattern, 13% isolated strains were completely sensitive to first-line drugs. 53% strains were MDR, 10% monodrug-resistant, and 7% polydrug-resistant. There was no significant association between DST pattern and outcome (P = 0.13). The irregular regimen was associated with MDR TB as twice as CAT I regimen failure (69.2% versus 35.4%, P = 0.004). Patients with MDR TB significantly experienced more side effects than non-MDR-TBs (47% versus 27%, P = 0.102). Of 100 patients, 72% were cured, 5% abandoned treatment, 12% died, 6% were classified as treatment failures, 1% relapsed, and 5% were transferred out. Of 53 patients with MDR TB, 33 subjects were cured and seven died. All together, successful outcome was achieved in 62.2%, 76%, and 76% of MDR TB, non-MDR TB, and completely sensitive cases, respectively. A retreatment strategy based on DST and replacing the Category II regimen with an intermediate regimen called revised CAT II may improve clinical outcomes among Category I treatment failures and defaults who found to have active, infectious MDR TB. This strategy significantly reduces delays to MDR TB diagnosis and to the initiation of MDR TB therapy. Success rate of this strategy is 62.2% and 72% in MDR TB and overall CAT I failure cases and defaulters, respectively.