Nosocomial spreading of extensive drug-resistant Acinetobacter baumannii (EDRAB) is an emerging problem. To clarify the association between prior antibiotic usage and subsequent EDRAB acquisition, we conducted a one-to-one matched case-control study among patients in all intensive care units (ICUs) at the National Taiwan University Hospital, Taipei, Taiwan, during a 1-year period. A total of 113 pairs of patients were identified. We measured prior antibiotics exposure in 4 perspectives: usage, overall treatment duration, accumulated dosage, and treatment potency. We found positive associations between EDRAB acquisition and prior usage of imipenem and meropenem across 4 measures, especially in usage and average treatment potency (usage, odds ratio [OR] imipenem = 3.7 with 95% confidence interval [CI] = 1.2–11.0, OR meropenem = 5.4 with 95% CI = 1.2–20.0; average treatment potency, OR imipenem = 5.3 with 95% CI = 1.3–22.0, OR meropenem = 3.4 with 95% CI = 1.0–12.0). Ceftazidime use with stronger treatment potency was also strongly associated with subsequent nosocomial EDRAB acquisition (OR = 5.5, 95% CI = 1.5–21.0). The OR of EDRAB acquisition greatly increased in patients who had previously been exposed to any 1 (OR = 5.5, 95% CI = 2.3–13.2) or to any 2 or 3 (OR = 11.1, 95% CI = 2.7–46.4) of the abovementioned antibiotics. Based on these findings, we conclude that usage of imipenem, meropenem, and/or ceftazidime is associated with subsequent acquisition of EDRAB in critically ill patients in ICUs.