The three-dimensional distortions of implant frameworks of two designs, reflecting the early "Toronto" U-shaped design and a current L-shaped design, were measured. A conventional commercial laboratory one-piece casting method, using a silver-palladium alloy, was used. Distortion was defined by five displacement variables for the centroids of each of five gold cylinders incorporated in each casting: three orthogonal translational displacements (dx, dy, and dz) and two rotational displacements (d theta x and d theta y). A computerized coordinate measuring machine was used. Overall translational displacements measured were -17.4 +/- 15 microns for dx, 6.1 +/- 7.1 microns for dy, and -0.1 +/- 1.7 microns for dz. The overall global translational displacement, dR, was 20.2 +/- 14.5 microns. Overall rotational displacements were -0.098 +/- 0.109 degrees for d theta x and -0.020 +/- 0.097 degrees for d theta y, giving an average absolute angular change, d theta R, of 0.158 +/- 0.077 degrees. No significant differences were found between U- and L-shaped castings, except for dy values for cylinders 2, 3, and 4, and d theta x values for cylinder 1. Cylinder location within the arch was associated with a specific direction and magnitude of translational and rotational displacement. Even small rotational displacements may manifest large gap appearances with the one-screw test because of the "moment arm" effect. However, other rotational displacements may be hidden, depending on the direction of tilt and the moment arm length.