Accurate impression making is an essential prerequisite for achieving a passive fit between the implant and the superstructure. The aim of this in vitro study was to compare the three-dimensional accuracy of open-tray and three closed-tray impression techniques.Three acrylic resin mandibular master models with four parallel implants were used: Biohorizons (BIO), Straumann tissue-level (STL), and Straumann bone-level (SBL). Forty-two putty/wash polyvinyl siloxane impressions of the models were made using open-tray and closed-tray techniques. Closed-tray impressions were made using snap-on (STL model), transfer coping (TC) (BIO model) and TC plus plastic cap (TC-Cap) (SBL model). The impressions were poured with type IV stone, and the positional accuracy of the implant analog heads in each dimension (x, y and z axes), and the linear displacement (ΔR) were evaluated using a coordinate measuring machine. Data were analyzed using ANOVA and post-hoc Tukey tests (α = 0.05).The ΔR values of the snap-on technique were significantly lower than those of TC and TC-Cap techniques (P < 0.001). No significant differences were found between closed and open impression techniques for STL in Δx, Δy, Δz and ΔR values (P = 0.444, P = 0.181, P = 0.835 and P = 0.911, respectively).Considering the limitations of this study, the snap-on implant-level impression technique resulted in more three-dimensional accuracy than TC and TC-Cap, but it was similar to the open-tray technique.