Purpose: The goal of this study was to foresee the fatigue life of two implant connections, evaluate the failure probability with several bone levels, and compare the in vitro test results with finite element results. Materials and Methods: Mechanical tests were done with 60 implants (Ø3.50 mm), and abutments were used. These implants were divided into two groups with 30 implants each: internal hexagon and Morse taper. Three bone levels and 10 implants for each level were analyzed. The first level was considered at the platform level, the second at 3 mm, and the last level at 5 mm above the platform resin. A quasi-static loading at 30 degrees was applied to the axis of the implant in a universal machine. Six models were created and assembled to reproduce the conditions used in the laboratory testing. All models had restricted all displacement at the bone (bottom and lateral). Loads employed in the numerical test were obtained experimentally. Loads and material properties were supposed to be random. Then, failure probability was calculated by the probabilistic methodology. Results: The internal hexagon group obtained the following mean fracture strengths: 2,092 N at the first level, 1,041 N at the second level, and 898 N at the third level. The mean fracture strengths for the Morse taper group were as follows: 1,687 N at the first level, 1,644 N at the second level, and 1,159 N at the third level. Results obtained by the finite element analysis are in accordance with the in vitro mechanical test results. The Morse taper group obtained a better behavior at bone levels 2 and 3 than the internal hexagon group. An important dependency between failure probability and bone level was found in the internal hexagon group. However, a similar behavior in levels 2 and 3 was obtained for the Morse taper group. Conclusion: In view of the mechanical results, the Morse taper group has a better behavior in bone levels 2 and 3 than the internal hexagon group. This is also in accordance with the probabilistic fatigue outcomes. Keywords: abutment, bone level, dental implant, failure probability, fracture strength, probabilistic approach