In addition to the pure measurement of bone mineral density (BMD) in osteodensitometry, the investigation of bone structure is becoming increasingly important for estimating fracture risk. In a clinical study, a risk score was proposed which separately assesses BMD and structural parameters for spongious and cortical bone and aggregates them into a single diagnostic parameter. In 120 lumbar vertebrae from 40 patients, BMD was determined separately for spongious and cortical bone by means of quantitative CT using a single energy procedure (SE-QCT/85 kV). In addition, structural parameters based on high resolution CT were calculated for the spongiosa and cortical bone. For all patients the number of osteoporosis-related fractures was determined on the entire skeletal system. According to WHO criteria, the patients were subdivided into four groups: 1, normal; 2, osteopenic; 3, osteoporotic without fractures; 4, severely osteoporotic. Weighting factors were determined by means of multivariate least-squares analysis and used to calculate a risk score of all parameters. The ability of the individual parameters and of the sum of discriminate between the individual groups was tested. If one considers the individual parameters (BMD and the fractal structural values for spongious and cortical bone), they allow a statistically significant separation of the four groups, although there is overlapping in the value ranges. In patients with fractures, there was a significant reduction in the cortical mineral density, accompanied by a deterioration in structural properties. The following individual values were obtained (minimum-mean-maximum): spongiosa BMD (mg ml-1), unfractured: 62-112-163, fractured: 9-48-77; cortical BMD (mg ml-1), unfractured: 190-287-405, fractured: 133-191-269; spongiosa structural parameter, unfractured: 0.35-0.73-1.01, fractured: 0.95-1.24-1.58; cortical structural parameter, unfractured: 18-31-65, fractured: 21-44-66. Above 77 mg ml-1 CaHA in the spongiosa and 270 mg ml-1 CaHA in cortical bone, no fractures were observed. By appropriately selecting the weighting factors, the score is free of overlapping between the groups with and without fractures (values: unfractured 1-9-15, fractured 16-21-29). With higher score values, the fracture risk is increasing.