OBJECTIVES: To study (1). changes in anthropometrics, joint range of motion (ROM), muscle strength, functional ability, caregiver assistance, and level of ambulation in children with osteogenesis imperfecta (OI) and (2). the prediction of clinical characteristics at the level of ambulation at follow-up and the prediction of clinical characteristics on progression or regression at the level of ambulation over time. DESIGN: Prospective study with follow-up of 4 years. SETTING: A children's hospital that serves a nationwide center for treatment and research in children with OI in the Netherlands. PARTICIPANTS: At follow-up, 49 children (24 boys, 25 girls; mean age +/- standard deviation, 11.3+/-3.8y; range, 5.2-19.4y) participated. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Anthropometry, joint ROM, muscle strength, fracture frequency, intramedullary rodding, level of ambulation, functional ability, and caregiver assistance. RESULTS: In type I OI, total joint ROM decreased significantly over time, especially in the lower extremities, with a significant decrease in generalized joint hypermobility according to Bulbena (median start, 7.5; interquartile range [IQR], 4-9; median end, 6; IQR, 2-7; P<.001). In types III and IV, a severe decrease in total joint ROM was present without significant changes over time. No significant changes in total muscle strength (upper or lower extremities) in the different types of OI were measured at follow-up. In OI type I, a significant increase in self-care (P=.003) and social function (P=.008) was measured; in type III, a significant increase in self-care (P=.003), mobility (P=.004), and social function (P=.005) was measured, with a significant decrease in parental assistance in self-care (P=.02) and mobility (P=.005). In type IV, a significant increase was observed in the self-care (P=.01) and social function domains (P=.02). Type of OI (regression coefficient=-1.0; 95% confidence interval [CI], -1.64 to -0.47) and total muscle strength were the only significant predictors for level of ambulation (regression coefficient=.01; 95% CI,.17-.32). Body weight was significantly lower in the group that progressed in level of ambulation (P=.03), whereas children with a decline in level of ambulation had significantly higher body weight (P=.05). CONCLUSIONS: Ours is the first study with a long-term follow-up that provides information concerning the natural course of developmental outcome parameters of OI in childhood. Joint ROM and muscle strength did not change significantly over time, possibly because of the biomechanical skeletal properties of the different OI types. Functional ability improved significantly over time, but, especially in types III and IV, did not reach normative values, possibly because of a plateau phase in functional ability. Knowledge of the natural course of the disease is essential to interpret the results from intervention studies.