This retrospective study evaluated the survival rate and the clinical outcome of an endosseous implant system, as well as the marginal bone level (MBL) and the impact of the machined collar neck position on crestal bone level changes up to 2 years later. A total of 96 implants were placed in 57 patients and loaded with a final restoration after at least 3 months of healing: 15 implants were immediately placed into fresh extraction sockets, and 81 were placed in healed ridges. Marginal bone loss around implants was determined radiographically at 12 to 24 months postloading. The effects on marginal bone loss of the site (mandible vs maxilla), location (anterior vs posterior), immediate vs delayed implant, smoking habit, sex, implant length and diameter, prosthesis type (screw-retained/cemented), and the machined collar position were evaluated. The implant survival rate was 98.96% at 1 year. The mean MBL decreased significantly (0.238 mm) between baseline and loading (P < .001; post hoc test) and decreased by 0.154 mm between loading and 1 year, but this was not significant (P = .085; post hoc tests). After 2 years, the mean MBL decreased significantly (0.263 mm) between baseline and loading (P < .001) and decreased by 0.111 mm between loading and 1 year and by 0.199 mm between loading and 2 years, but these were not significant (P > .05; post hoc tests). The mean bone loss after 1 and 2 years was not significantly associated with implant type or site, smoking habit, or type of implant surgery (P = .792). However, the mean bone loss was significantly associated with the type of prosthesis and was significantly greater for cemented prostheses compared to screw-retained prostheses. A supracrestal position of the machined collar (tissue level) was associated with no bone loss, while placing the collar below the alveolar crest led to bone loss over 2 years. Of the factors evaluated, marginal bone loss after 1 and 2 years was significantly associated with prosthesis type and the machined collar position.