The objective of this study was to statistically correlate in vivo measurements of prosthesis misfit and change of marginal bone level in implants placed in the edentulous maxilla. Two groups, each comprising seven patients, were followed up either prospectively for 1 year or retrospectively for the last 4 years of the 5-year period after second-stage surgery. Measurements of prosthesis misfit were performed by means of a three-dimensional photogrammetric technique, and marginal bone levels were measured from standard intraoral radiographs. Results showed that none of the prostheses presented a completely passive fit to the implants in vivo. Furthermore, similar distortions of the prostheses were found in the two groups, indicating that the implants seem to be stable and do not move, even after several years in function. The maximal range of three-dimensional distortion of cylinder center points was about 275 µm for both groups. Mean center point misfit was 111 (SD 59) and 91 (SD 51) µm for the 1-year and 5-year groups, respectively. The corresponding mean marginal bone loss was 0.5 and 0.2 mm for the two follow-up groups. No statistical correlations (P > .05) between change of marginal bone levels and different parameters of prosthesis misfit were observed in the two groups. The study indicated that a certain biologic tolerance for misfit may be present. The degree of misfit reported in the study was clinically acceptable with regard to observed marginal bone loss. Keywords: bone loss, complication, follow-up, framework, implant, misfit, precision