Physical exercise has a favourable impact on bones, but optimum training strategies are still under discussion. In this study we compared the effect of slow and fast resistance exercises on various osteodensitometric parameters. 53 postmenopausal women were randomly assigned to a strength (ST) or a power training group (PT). Both groups carried out a progressive resistance training, a gymnastics session, and a home training over a period of 12 months. During the resistance training the ST group used slow and the PT group fast movements, otherwise there were no training differences. All subjects were supplemented with Ca/Vit-D. At baseline and after 12 months, BMD was measured at the lumbar spine, proximal femur, and distal forearm by DXA. We also measured anthropometric data and maximum static strength. Frequency and grade of pain were assessed by questionnaire. After 12 months, significant between-group differences were observed for BMD at the lumbar spine (L1-L4) (p<0.05) and the total hip (p<0.05). While the PT group maintained BMD at the spine (+0.7±2.1%, n.s.) and the total hip (0.0±1.7%, n.s.), the ST group lost significantly at both sites (spine: -0.9±1.9%; p<0.05; total hip: -1.2±1.5%; p<0.01). No significant between-group differences were observed for anthropometric data, maximum strength, BMD of the forearm, or frequency/grade of pain. These findings suggest that power training is more effective than strength training in reducing bone loss in postmenopausal women.