Background: According to literature, peri-implant bone loss is minimized on implants with microthreaded neck design and internal type of abutment connection. However, most clinical studies may be biased due to confounding factors.Purpose: This nonblinded RCT assessed the effect of implant neck (microthreaded vs non-microthreaded) as well as the type of abutment connection (internal conical vs external flat-to-flat) on peri-implant bone stability and peri-implant health after at least 36 months.Materials and methods: Twenty-five patients were treated with a maxillary implantsupported bar-retained overdenture on four different implant types: internal connection with microthreads (I-MT), internal connection without microthreads (I-NMT), external connection with microthreads (E-MT), and external connection without microthreads (E-NMT). To control confounding factors, all other design features were similar. A linear mixed-model analysis or mixed-model logistic regression analysis was used to determine the effect of implant type on bone level, probing pocket depth, bleeding on probing, and plaque.Results: Four out of 98 implants (4.1%) placed in 25 patients failed during provisionalization and were replaced. Mean overall bone loss after 6 months was 0.39 mm (SD 0.62, range 0.00-3.48) with limited additional bone loss of 0.04 mm (SD 0.54, range À1.80-1.63) after at least 3 years. Microthreads or connection type had no effect on the bone level, probing pocket depth, bleeding on probing, nor plaque.Conclusions: With 96% of implant survival, the maxillary overdenture supported with a bar on four implants yield a predictable outcome and the implant-abutment connection type (internal vs external) and implant neck design (microthreaded vs nonmicrothreaded) have no influence on peri-implant bone remodeling after initial bone remodeling nor up to 4 years of function. Peri-implant bone levels are within