Objectives Lumbar lordosis (LL) is a key adaptation for bipedalism, but factors underlying curvature variations remain unclear. This study tests three biomechanical models to explain LL variability. Materials and methods Thirty adults (15 male, 15 female) were scanned using magnetic resonance imaging (MRI), a standing posture analysis was conducted, and lumbar range of motion (ROM) was assessed. Three measures of LL were compared. The trunk's center of mass was estimated from external markers to calculate hip moments (Mhip) and lumbar flexion moments. Cross‐sectional areas of lumbar vertebral bodies and trunk muscles were measured from scans. Regression models tested associations between LL and the Mhip moment arm, a beam bending model, and an interaction between relative trunk strength (RTS) and ROM. Results Hip moments were not associated with LL. Beam bending was moderately predictive of standing but not supine LL (R2 = 0.25). Stronger backs and increased ROM were associated with greater LL, especially when standing (R2 = 0.65). The strength‐flexibility model demonstrates the differential influence of RTS depending on ROM: individuals with high ROM exhibited the most LL variation with RTS, while those with low ROM showed reduced LL regardless of RTS. Discussion Hip moments appear constrained suggesting the possibility of selection, and the beam model explains some LL variability due to variations in trunk geometry. The strength‐flexibility interaction best predicted LL, suggesting a tradeoff in which ROM limits the effects of back strength on LL. The strength‐flexibility model may have clinical relevance for spinal alignment and pathology. This model may also suggest that straight‐backed Neanderthals had reduced lumbar mobility.