To assess the association between protein intakes and bone mass accrual in girls, data were analysed for 757 pre-pubertal girls (mean age 10·1 years) in urban Beijing, China, who participated in a 5-year study including 2 years of milk supplementation (intervention groups only) and 3 years of follow-up study. At 0, 12, 24, 48 and 60 months from the baseline, bone mass of the proximal or distal forearm (PF or DF) and total body (TB) was measured with dual energy X-ray absorptiometry; dietary intakes were assessed by a 3-d food record (including two weekdays and one weekend day). Linear mixed models were used and continuous variables were logarithm transformed. The mean longitudinal Ca intake (432–675 mg/d on average) positively influenced bone mineral content (BMC) at TB, PF and DF after controlling for baseline bone mass and other possible confounders. However, negative associations were observed between protein intake (55·9–61·0 g/d on average) and BMC accrual at TB, PF or DF (β = − 1·92, − 10·2 or − 4·82, respectively, P < 0·01) after adjustment. When protein intake was considered according to animal or plant food sources, protein from animal foods, particularly meat, had significant negative effects on BMC accrual at DF or PF after adjustment. It was concluded that higher protein intake, especially from animal foods, appeared to have a negative effect on bone mass accrual in Chinese pubertal girls with low Ca intakes.