The purpose of this study was to evaluate crestal bone loss around 282 two-piece implants with straight (n = 193) and platform-switched (n = 89) abutment connections after placement at various crestal levels. Implants were assigned into two groups according to straight and platform-switched abutment connections. Each group was further subdivided into three groups depending on the location (supracrestal, crestal, or subcrestal) of the implant cervical platform. Linear measurements of bone resorption were made from the implant's platform to the first point of bone-to-implant contact at the time of implant placement and 2 years postrestoration. Data were statistically analyzed. Statistically significant differences were found between subgroups in both straight and platform-switched categories. The only nonstatistically significant difference (P = .341) arose when comparing the supra- and subcrestal locations in the straight abutment connection group. The platform-switched group exhibited significantly less bone loss (P = .046) only in subcrestal locations. The platform-switched concept was not beneficial during the overall comparison, but it was for the subcrestal location of the abutment connection. Crestal placement of the implant-abutment connection resulted in higher marginal bone resorption in both straight and platform-switched abutments.