The accuracy of impressions that transfer the relationship of the implant to the metal framework of the prosthesis continues to be a problem. This study was designed to evaluate the accuracy of the transfer process under variable conditions with regard to implant analog angulations, impression materials, and techniques. Replicas (n = 60) of a metal matrix (control) containing four implants at 90 degrees , 80 degrees , 75 degrees , and 65 degrees in relation to the horizontal surface were obtained by using three impression techniques: T1-indirect technique with conical copings in closed trays; T2-direct technique with square copings in open trays; and T3-square copings splinted with autopolymerizing acrylic resin; and four elastomers: "P"-polysulfide; "I"-polyether; "A"-addition silicone; and "Z"-condensation silicone. The values of the implant analog angulations were assessed by a profilometer to the nearest 0.017 degrees , then submitted to analysis of variance for comparisons at significance of 5% (P < .05). For implant analog at 90 degrees , the material "A" associated with T2 and material "Z" with T3 behaved differently (P < .05) from all groups. At 80 degrees , all materials behaved differently (P < .01) with T1. At 75 degrees , when T1 was associated, materials "P" and "A" showed similar behavior, as well as materials "I" and "Z"; however, "P" and "A" were different from "I" and "Z" (P < .01). When T3 was associated, all experimental groups behaved differently among them (P < .01). At 65 degrees , the materials "P" and "Z" behaved differently (P < .01) from the control group with T1, T2, and T3; the materials "I" and "A" behaved differently from the control group (P < .01) when T1 and T2, respectively, were associated. The more perpendicular the implant analog angulation is in relation to the horizontal surface, the more accurate the impression. The best materials were material "I" and "A" and the most satisfactory technique was technique 3.