OBJECTIVES:To assess the overall prescribing and chronic use of benzodiazepines (BZs) in nursing homes before and after implementation of BZ guidelines by the Health Care Finance Administration (HCFA). A second aim was to identify facility predictors of BZ prescribing and chronic use after guideline implementation. DESIGN: Nonexperimental, observational study design. SETTING: Drug use was assessed in 16 randomly selected skilled nursing facilities in Wisconsin before 1990Wisconsin before and again in 1993Wisconsin before -1994. PARTICIPANTS: One thousand one hundred eighty-one Medicaid-funded residents at baseline and 1,650 Medicaid-funded residents at follow-up. MEASUREMENTS: Drug use measures included percentage of residents with prescribed BZs and percentage of residents with chronic BZ use (defined as Ն 4 months of continuous use). A 4-month cutoff was used because HCFA guidelines consider this length of continuous use to be potentially excessive. Facility predictors of drug use included resident demographic and diagnostic mix, size, ownership, Medicaid per diem, and nurse staffing. RESULTS: Before 1990, one-fourth of all residents were prescribed a BZ and nearly one-tenth of all residents had chronic BZ use. Federal guidelines had nonsignificant effects on these measures; BZ prescribing declined only 3.6% (26.4-22.8%) and chronic use declined only 1.3% (9.2-7.9%). Regression results showed that overall prescribing at follow-up was higher in facilities with more female residents (  ϭ .52, P ϭ .018) and residents with psychiatric disorders (  ϭ .41, P ϭ .045). However, nurse staffing was the only significant predictor of chronic BZ use. As expected, better nurse staffing was associated with lower rates of chronic use (  ϭ Ϫ 0.46, P ϭ .037). CONCLUSION: Efforts to improve BZ use in nursing homes must address the need for better nurse staffing and better assessment and management of psychiatric disorders that can trigger BZ prescribing.