Nursing home residents are likely to differ from community older adults when their gait parameters are compared, as nursing home residents present more falls. The study aim was to identify the main fall (occurrence predictors) (anthropometrics, functional and gait-related parameters) between older adults living in community and nursing homes during self-selected (SSWS) and fast walking speeds (FWS). A hundred and sixty-five older adults were selected from the community (n = 92) and nursing home (n = 73) with and without fall history. They were assessed for fall history, functionality, cognitive status, and several gait parameters in SSWS and FWS conditions. Fallers differed from non-fallers in the SSWS, while such differences were not evidenced during the FWS. Cadence and stride width did not differ when living backgrounds were compared. Nursing home residents walked slower than their non-institutionalized peers, regardless of fall history or walking speed. Besides, binary logistic regression analysis showed that living in a nursing home, age, body mass index (BMI), mini-mental state examination (MMSE), and step width were related to falls in the SSWS. On the other hand, living in a nursing home, having a larger BMI and having low MMSE scores were fall predictors in the FWS. Fall occurrence can be identified by factors related to living in nursing homes, cognitive status, BMI, and gait parameters, at the SSWS. Cognitive status and BMI are related to falls in the FWS for those living in nursing homes.