Background: The aim of this study was to compare the accuracy of different direct implant impression techniques for edentulous arches with multiple implants. Methods: Five experimental groups (n = 5) were assembled. Experimental models were created by a direct splinted technique (EG2 to EG5) and a non-splinted technique (EG1). In EG2 and EG3 synOcta impression copings were splinted with an acrylic resin bar, and in EG4 and EG5 with a light-curing composite resin bar. In EG3 and EG5 the resin bars were sectioned, while the other experimental groups were not. Three-dimensional discrepancies were measured by a computerized coordinate measuring machine. Distortion values among the groups were analysed using one-way repeated measures ANOVA. The post hoc Tukeys test was then performed for multiple comparisons. Results: The highest accuracy was obtained in EG2 (mean deviation: 12.70 mu m). The acrylic bars demonstrated less deviation (12.70 mu m and 22.71 mu m) from the master model than the light-curing composite resin groups and the non-splinted group (41.09 mu m). The post hoc Tukeys test showed no significant difference among the groups when the effect of splint design on accuracy was investigated. Conclusions: For situations where impressions of multiple implants are to be made, splinting impression copings with acrylic resin demonstrate superior results than the non-splinted technique and splinting with light-curing composite.