Background: Perinatally acquired HIV infection commonly causes stunting in children, but how this affects bone and muscle development is unclear. We investigated differences in bone and muscle mass and muscle function between children with HIV (CWH) and uninfected children. Setting: Cross-sectional study of CWH (6 to 16 years) receiving antiretroviral therapy (ART) for >6 months and children in the same age-group testing HIV-negative at primary health clinics in Zimbabwe. Methods: From Dual-energy X-ray Absorptiometry (DXA) we calculated total-body less-head (TBLH) Bone Mineral Content (BMC) for lean mass adjusted-for-height (TBLH-BMCLBM) Z-scores, and lumbar spine (LS) Bone Mineral Apparent Density (BMAD) Z-scores. Results: The 97 CWH were older (mean age 12.7 vs. 10.0 years) and therefore taller (mean height 142cm vs. 134cm) than those 77 uninfected. However, stunting (height-for-age Z-score = -2) was more prevalent in CWH (35% vs. 5%, p <0.001). Amongst CWH, 15% had low LS-BMAD (Z-score = -2) and 13% had low TBLH-BMCLBM, vs. 1% and 3% respectively in those uninfected (both p = 0.02). After age, sex, height and puberty adjustment, LS-BMAD was 0.33 SDs (95%CI -0.01, 0.67; p=0.06) lower in CWH, with no differences in TBLH-BMCLBM, lean mass or grip strength by HIV status. However, there was a strong relationship between age at ART initiation and both LS-BMAD Z-score (r=-0.33, p=0.001) and TBLH-BMCLBM Z-score (r=-0.23, p=0.027); for each year ART initiation was delayed a 0.13 SD reduction in LS-BMAD was seen. Conclusion: Size-adjusted low bone density is common in CWH. Delay in initiating ART adversely affects bone density. Findings support immediate ART initiation at HIV diagnosis.