This study describes a clinically oriented anatomical assessment of anteversion on 32 cadavers ranging from 61 to 89 years. The method used a three-dimensional reference system based on functional axes of the femur. Each soft tissue-free femur was mounted on an osteometric table and aligned to its functional axes. The long axes were defined as passing from the centre of the femoral head to the femoral attachment of the posterior cruciate ligament (PCL). A line that ran through the PCL attachment (equal distal origin of this system) and was parallel to a transepicondylar line served as transverse axis. Anteversion of the femur was defined as an angle formed to the transverse axis by a line running through the centre of the femoral head through the midpoint of the narrowest segment of the femoral neck. The measurement mean for anteversion among these specimens was 7.4 degrees with a range from -10.8 degrees (retroversion) to 22.1 degrees. There was no statistical difference in mean values (p less than 0.05) between sexes or between right and left sides of the group; however, there were large variations for anteversions when each side in the same individual was compared (although there was no dominant side). Retroversions were observed in four of 32 femurs (12.5%). No correlation was found between the anteversion in these femurs and rotational geometry at the knee. We compared our data with those obtained by conventional techniques, by which anteversion for each femur was measured after the bone had been placed on a flat surface.(ABSTRACT TRUNCATED AT 250 WORDS)