Human movement analysis using stereophotogrammetry is based on the reconstruction of the instantaneous laboratory position of selected bony anatomical landmarks (AL). For this purpose, knowledge of an AL's position in relevant bone-embedded frames is required. Because ALs are not points but relatively large and curved areas, their identification by palpation or other means is subject to both intra- and inter-examiner variability. In addition, the local position of ALs, as reconstructed using an ad hoc experimental procedure (AL calibration), is affected by photogrammetric errors. The intra- and inter-examiner precision with which local positions of pelvis and lower limb palpable bony ALs can be identified and reconstructed were experimentally assessed. Six examiners and two subjects participated in the study. Intra- and inter-examiner precision (RMS distance from the mean position) resulted in the range 6-21 mm and 13-25 mm, respectively. Propagation of the imprecision of ALs to the orientation of bone-embedded anatomical frames and to hip, knee and ankle joint angles was assessed. Results showed that this imprecision may cause distortion in joint angle against time functions to the extent that information relative to angular movements in the range of 10 degrees or lower may be concealed. Bone geometry parameters estimated using the same data showed that the relevant precision does not allow for reliable bone geometry description. These findings, together with those relative to skin movement artefacts reported elsewhere, assist the human movement analyst's consciousness of the possible limitations involved in 3D movement analysis using stereophotogrammetry and call for improvements of the relevant experimental protocols.