To compare the effects of high-intensity resistance and impact training (HiRIT) to low-intensity, Pilates-based exercise (LiPBE) on proximal femur geometry and explore the influence of antiresorptive medication on those effects. Postmenopausal women with low bone mass, on or off antiresorptive bone medications were randomly allocated, stratified on medication intake, to eight months of twice-weekly, supervised HiRIT (Onero™) or LiPBE (Buff Bones®). 3D hip software was used to analyse proximal femur DXA scans. Outcomes included femoral neck (FN) and total hip (TH), volumetric (e.g. vBMC, vBMD) and geometric (e.g. cortical thickness, cross-sectional area [CSA], section modulus [Z]) indices of bone strength. Data were analysed using analysis of variance. Scans of 102 women were examined: LiPBE, 43; HiRIT, 37; LiPBE-med, 11; HiRIT-med, 11. HiRIT improved TH trabecular vBMC and vBMD (3.1 ± 1.1% versus − 1.2 ± 1.2%, p = 0.008; and 1.5 ± 1.0% versus − 1.6 ± 1.2%, p = 0.042, respectively) and FN and TH total vBMC (2.0 ± 0.8% versus − 0.2 ± 0.7%, p = 0.032; and 0.7 ± 0.4% versus − 0.8 ± 0.6%, p = 0.032, respectively), compared to losses in LiPBE. HiRIT also increased Z while LiPBE did not (p = 0.035). The combination of HiRIT and medication achieved greater improvements in FN total and trabecular vBMD, total BMC, CSA and Z than HiRIT alone. HiRIT improved geometric parameters of proximal femur strength, while LiPBE exercise was largely ineffective. Medication may enhance some HiRIT effects. Findings suggest reduced hip fracture risk in response to HiRIT. Trial registration number ACTRN12617001511325.