Purpose: The aim of this study was to evaluate, using scanning electron microscopy (SEM), the accuracy of four impression techniques for osseointegrated implants (with or without acrylic resin splinting and with irreversible hydrocolloid or polyvinyl siloxane [PVS] impression material). Materials and Methods: A metal master model was made with three implant analogs and two prosthetic spaces. This model was used as the standard for all impressions. Two impression materials were used (irreversible hydrocolloid and PVS) and two transfer techniques were used (squared impression copings indexed by the impression material and squared impression copings splinted with acrylic resin). Four groups were therefore analyzed (n = 5): IH = irreversible hydrocolloid only, IHS = irreversible hydrocolloid + splint, P = PVS only, and PS = PVS + splint. A reference framework made with palladium-silver alloy over the UCLA abutment was created on the master model. The fit of this structure to the master model was used as a reference. SEM images of the front and side gaps between the abutments and the implant analogs were created and then measured using image analysis software. Results: IH presented the largest misfit. The splinted impression copings generated a smaller marginal gap than the indexed material technique, irrespective of the impression material used. There was no significant difference between IHS, P, PS, and the reference (multivariate test, Wilks criteria). However, PS presented a standard deviation that was three times lower than those of the other groups, and its mean was closer to the reference. Conclusions: The IH impression technique was the least accurate technique. There was no difference between IHS, P, and PS techniques with regard to the reference constant. The impression techniques that used splinted impression copings generated more accurate casts, irrespective of the impression material. Palabras clave: dental implants, impression, irreversible hydrocolloid, polyvinyl siloxane, splinting