Dose range studies for polymyxin B (PMB) regimens of 0.75 to 12 mg/kg given every 12 h (q12h) were evaluated for bacterial killing and resistance prevention against an AmpC-overexpressing Pseudomonas aeruginosa and a blaKPC-3-harboring Klebsiella pneumoniae in 10-day in vitro hollow-fiber models. An exposure-response was observed. But all regimens failed due to regrowth. Lower-dose regimens amplified isolates that expressed transient, lower-level adaptive resistance to PMB (MICs ≤ 4 mg/liter). Higher PMB dosages amplified isolates that expressed this resistance mechanism, a higher-MIC "moderately stable" adaptive resistance, and a higher-MIC stable resistance to PMB. Failure of the highest dose regimens was solely due to subpopulations that expressed the two higher-level resistances. Total and bioactive PMB concentrations in broth declined below targeted PK profiles within hours of treatment initiation and prior to bacterial regrowth. With treatment failure, the total PMB measured in bacteria was substantially higher than in broth. But the bioactive PMB in broth and bacteria were low to nondetectable. Together, these findings suggest a sequence of events for treatment failure of the clinical regimen. First, PMB concentrations in broth are diluted as PMB binds to bacteria, resulting in total and bioactive PMB in broth that is lower than targeted. Bacterial regrowth and treatment failure follow, with emergence of subpopulations that express transient lower-level adaptive resistance to PMB and possibly higher-level adaptive and stable resistances. Higher-dose PMB regimens can prevent the emergence of transient lower-level adaptive resistance, but they do not prevent treatment failure due to isolates that express higher-level resistance mechanisms.