This study aimed to analyze the accuracy resulting from dual-arch impressions when compared to conventional impressions in complex preparations (ie, inlay and partial crown). One hundred eighty impressions were made using two different dual-arch trays; conventional trays served as the control. The accuracy of the dies obtained (Fuji-Rock EP, GC Europe) was assessed indirectly from the change of 59 transversal dimensions. Statistical analysis (t test, analysis of variance) revealed that less rigid dual-arch trays performed better than rigid ones. Though the inlay preparation was more difficult to reproduce with dual-arch trays, it can be concluded that the accuracy obtainable with nonrigid dual-arch trays is comparable to impressions taken from full-arch trays.