Bone-mineral density was studied in a heterogeneous group of 139 children (mean age, nine years; range, three to fifteen years) who had spastic cerebral palsy. The evaluation included serum analyses and a nutritional assessment based on a dietary history and anthropometric measurements. The bone-mineral density of the proximal parts of the femora and the lumbar spine was measured with dual-energy x-ray absorptiometry and was normalized for age against a series of ninety-five normal children and adolescents who served as controls. Bone-mineral density varied greatly but averaged nearly one standard deviation below the age-matched normal means for both the proximal parts of the femora (-0.92 standard deviation) and the lumbar spine (-0.80 standard deviation). Ambulatory status was the factor that best correlated with bone-mineral density. Nutritional status, assessed on the basis of caloric intake, skinfolds, and body-mass index, was the second most significant variable. The pattern of involvement, durations of immobilization in a cast, and a calcium intake of less than 500 milligrams per day were additional factors of less significance. The age when the child first walked, previous fractures, use of anticonvulsants, and serum vitamin-D levels did not correlate with bone-mineral density after adjustment for covariance with the ambulatory status and the nutritional status. Serum levels of calcium, phosphate, alkaline phosphatase, and osteocalcin were not reliable indicators of low bone-mineral density.