BACKGROUND: The scientific literature has shown that an association between polypharmacy and frailty exists. However, few studies have also considered drug interactions and the use of potentially inappropriate medications. OBJECTIVE: To evaluate the association between the use of drugs and frailty among community-dwelling older people. DESIGN AND SETTING: Cross-sectional study carried out among 580 older people in Uberaba (MG). METHODS: Data were collected at these older people's homes using instruments validated in Brazil. Descriptive, bivariate and binary logistic regression analyses were performed (P < 0.05). RESULTS: Most of these individuals were classified as pre-frail (55.7%), while 13.1% were frail. It was found that 31.7% of them presented polypharmacy, 41.7% had drug interactions and 43.8% were using potentially inappropriate medications. In the initial model, polypharmacy (odds ratio, OR = 1.91; confidence interval, CI = 1.27-2.86) and use of potentially inappropriate medications (OR = 2.45; CI = 1.68-3.57) increased the chance that these older people would be pre-frail or frail. In the final adjusted model, use of potentially inappropriate drugs remained associated with the outcome (OR = 2.26; CI = 1.43-3.57). CONCLUSION: Use of potentially inappropriate medications was the independent variable that explained the occurrence of frailty in a representative sample of community-dwelling older adults.