This study evaluated the effect of a modified tray design on the accuracy of implant impressions in comparison with the non-splinted and splinted impression techniques. Two titanium frameworks were produced to fit two parallel implants and two divergent implants with a 15 o angle. According to the frameworks employed, two acrylic resin master models were fabricated. For each model, 10 impressions were taken with every technique. The maximum framework principal strain was calculated for every generated cast. For the parallel implant model, the strains of the nonsplinted (118.4 με), splinted (89.0 με), and modified tray design impression (49.4 με) techniques were statistically similar (P = 0.16). For the divergent implant model, all the impression techniques showed a considerably higher strain than the parallel implant model. The splinted (287.0 με) and the modified (262.9 με) tray design impression techniques showed similar strains for the divergent implant model, which were significantly less than the strains for the non-splinted impression (518.0 με) technique (P < 0.05). Therefore, for two parallel implants, all the impression techniques exhibited similar accuracy. When angulation existed between the implants, the splinted and the modified tray design impression techniques were more accurate than the non-splinted impression technique.