The main purpose of the present study was to examine factors that affect bone mineral density (BMD) in female ballet dancers. Training history, Ca intake, body composition, total body BMD (TBMD) and site-specific BMD, and bone mineral content were described in twenty-four female ballet dancers (mean age 22·6 (SD 4·5) years). Training history was determined by questionnaires, Ca intake by 7 d dietary record, BMD and bone mineral content by dual-energy X-ray absorptiometry (DXA), total body water by 2H dilution, extracellular water by bromide dilution, body fat by underwater weighing (UWW; two-component model), DXA, and the four-component (4C) model. Dancers had a significantly lower body mass index (BMI 18·9 (SD 1·0) kg/m2) than controls (21·3 (SD 19) kg/m2), with significantly lower percentage body fat (17·4 (SD 3·9) v. 24·4 (SD 5·1)) but comparable fat-free mass. Mean TBMD (1·147 (SD 0 ·069) g/cm2) was significantly higher (6 %) compared with that of a reference population. These high values could be attributed to the high BMD of legs and pelvis, the weight-bearing sites of the dancer's body. No relationship was found between age, start of ballet classes, period (years) of dancing, Ca intake, and BMD (total and site-specific). However, TBMD was positively related to BMI, and negatively related to the age of menarche. BMD of the legs was significantly related to daily period (h) of training. Depending on the method used the percentage body fat ranged from 16·4 (by DXA) to 18·3 by the 4C model. These differences were significantly related to the TBMD. Percentage body fat by the different methods was not significantly different, except for DXA and 4C model. The present study showed that, despite the factors that have a negative effect on BMD, such as low body mass and late menarche, BMD in female ballet dancers was relatively high. These high values were probably caused by high levels of weight-bearing physical activity.