Over 95% of hip fractures in older adults are caused by falls. Understanding the circumstances of falls involving hip fracture should guide prevention efforts. This study examined falls captured on video in long-term care and compared the mechanics of falls that resulted in hip fracture to falls that did not. Between 2008 and 2017, we video-captured 1730 falls by 534 individuals. Hip fracture was documented in 34 falls (age=85±8 years, 65% female). We analysed each video with a structured questionnaire to determine the biomechanical characteristics of falls. Generalized Estimating Equation statistical models were used to calculate risk ratios (RRs) for hip fracture associated with each factor. All 34 falls involving hip fracture were from standing height and involved the pelvis striking the floor. Falls during walking or standing were more likely to result in hip fracture (RR=7.38; 95% CI: 2.27–24.02) than other activities. When comparing to other directions, hip fracture was more common in falls initially directed sideways (2.70; 1.39–5.21) and for sideways landing configurations (5.55; 2.61–11.78). Impact to the lateral aspect of pelvis created higher risk (3.34; 1.55–7.17) compared to an impact to the anterior or posterior pelvis. Among hip fracture cases, pelvis was among the first three sites of impact. Impact to pelvis occurred on the posterolateral aspect in 70% of cases, lateral in 18%, posterior in 6%, and anterolateral in 6%. Hip protectors were worn in 66% of cases that did not involve hip fracture and in 47% involving hip fracture, which showed a decreased risk for fractures (0.46; 0.23–0.95). In summary, the mechanics of falls involving hip fracture were different than non-fracture falls regarding fall height, fall direction, impact locations, and use of hip protectors. The high prevalence of impact to the posterolateral aspect of pelvis may be important to the design of wearable hip protectors.