Background: The concept of an ideal sitting posture is often used in practice but lacks a basis in evidence. Objective: We designed a cross-sectional, comparative, matched study to determine the effects of chair and posture on lumbar curvature in 10 patients with chronic non-specific low back pain (CLBP; mean pain duration 24 ± 18 months) and 10 healthy matched controls. Methods: Pelvic incidence, sacral slope and lumbar curvature were measured on computed radiographs by 2 blinded clinicians for subjects in 2 postures (upright vs slumped sitting) and on 2 chairs (usual flat chair vs kneeling chair). Results: The reliability of measures was excellent (intraclass correlation coefficient. >. 0.9). As hypothesized, the expected sacral slope and lumbar lordosis changed between standing and sitting on a kneeling chair as compared with a usual chair (P<. 0.0001) and less in patients than controls (P = 0.046) for lordosis only. In addition, as expected, changes were more pronounced with slumped than upright sitting (P<. 0.0001). An interaction between chairs and postures for lumbar lordosis (P = 0.02) indicated more pronounced effects of the chair in slumped sitting. Therefore, lumbar lordosis was reduced less when sitting on a kneeling chair as compared with a usual chair. Conclusions: Although healthy subjects showed more reduction in lordosis between standing and sitting, the chair effect was found in both CLBP patients and healthy subjects.