The objective of this paper is to present a method to describe the three-dimensional variations of the geometry of the three portions forming the distal part of the human femur: the medial and lateral femoral condyles and the intercondylar fossa. The contours of equally spaced sagittal slices were digitized on the distal femur to determine its surface topography. Data collection was performed using a digitizer system which utilizes low-frequency, magnetic field technology to determine the position and orientation of a magnetic field sensor in relation to a specified reference frame. The generalized reduced gradient optimization method was used to reconstruct the profile of each slice utilizing two primitives: straight-line segments and circular arcs. The profile of each slice within the medial femoral condyle was reconstructed using two circular arcs: posterior and distal. The profile of each slice within the lateral femoral condyle was reconstructed using three circular arcs: posterior, distal and anterior. Finally, the profile of each slice within the intercondylar fossa was reconstructed using two circular arcs: proximal-posterior and anterior, and a distal-posterior straight-line segment tangent to the proximal-posterior circular arc. Combining the data describing the profiles of the different slices forming the distal femur, the posterior portions of each of the medial and lateral femoral condyles were modelled using parts of spheres having an average radius of 20 mm. The anterior portion of the lateral condyle was approximated to a right cylinder having its circular base parallel to the sagittal plane with an average radius of 26 mm. The anterior portion of the intercondylar fossa was modelled using an oblique cylinder having its circular base parallel to the sagittal plane with an average radius of 22 mm. Furthermore, it is suggested that the distal portion of the lateral femoral condyle could be modelled using parts of two oblique cones while the distal portion of the medial femoral condyle could be modelled using a part of a single oblique cone, all cones having their circular bases parallel to the sagittal plane. It is also suggested that the posterior portion of the intercondylar fossa could be modelled using two oblique cones: a proximal cone having its base parallel to the sagittal plane and a distal cone having its base parallel to the frontal plane.