Objectives: This study examined the reliability by which connectors in a screw-retained prosthesis can maintain adequate joint preload compared to cemented implant prosthesis when each was subjected to simulated function. Methods: Six implant fixtures representing the mandible were set in caststone, three within each of two experimental platforms. In platform I, a three-unit fixed partial denture was constructed and attached to the three fixtures with screws. Torque value was assigned as an average value (14.32 Ncm, n=20) when maximum finger strength was tested, using a manual hand driver. Each screw was tightened using a dynamometer. In platform II a three-unit fixed partial denture was constructed and cemented to corresponding abutments, which were also tightened using the same dynamometer. Each prosthesis was then loaded with 4 kg at 25,000 cycles of vertical force to simulate three months of occlusal function. After completion of the cycles, the prostheses were removed and the backout torque for each platform was measured using the same dynanometer. Results: Upon removing the cemented implant prosthesis, back out torque was measured with the dynamometer for platform II (9.46 Ncm, +/- 0.12, n=3). In platform I the screw-retained implant prosthesis was removed and back out torque measured with the same instrument (7.56 Ncm, +/- 3.33, n=3). Conclusions: The small number of samples in this pilot study prohibits statistical comparisons. However, based on observations made in this study, more consistent torque maintenance reliability is suggested with cemented restorations. Considering industry standard failure levels of 0.6x (x=original torque), only one fastener failed to retain adequate tightness.