Over 95% of hip fractures in older adults are caused by falls, yet only 1% to 2% of falls result in hip fracture. Our current understanding of the types of falls that lead to hip fracture is based on reports by the faller or witness. We analyzed videos of real‐life falls in long‐term care to provide objective evidence on the factors that separate falls that result in hip fracture from falls that do not. Between 2007 and 2018, we video‐captured 2377 falls by 646 residents in two long‐term care facilities. Hip fracture was documented in 30 falls. We analyzed each video with a structured questionnaire, and used generalized estimating equations (GEEs) to determine relative risk ratios (RRs) for hip fracture associated with various fall characteristics. All hip fractures involved falls from standing height, and pelvis impact with the ground. After excluding falls from lower than standing height, risk for hip fracture was higher for sideways landing configurations (RR = 5.50;​ 95% CI, 2.36–12.78) than forward or backward, and for falls causing hip impact (3.38;​ 95% CI, 1.49–7.67). However, hip fracture risk was just as high in falls initially directed sideways as forward (1.14;​ 95% CI, 0.49–2.67), due to the tendency for rotation during descent. Falling while using a mobility aid was associated with lower fracture risk (0.30;​ 95% CI, 0.09–1.00). Seventy percent of hip fractures involved impact to the posterolateral aspect of the pelvis. Hip protectors were worn in 73% of falls, and hip fracture risk was lower in falls where hip protectors were worn (0.45;​ 95% CI, 0.21–0.99). Age and sex were not associated with fracture risk. There was no evidence of spontaneous fractures. In this first study of video‐captured falls causing hip fracture, we show that the biomechanics of falls involving hip fracture were different than nonfracture falls for fall height, fall direction, impact locations, and use of hip protectors.