Statement of problem. It has been suggested that articulated casts fabricated with the double-arch impression technique may have superior occlusal accuracy than those fabricated with a single complete-arch impression. However, lack of tray rigidity may lead to flexure of the impression/tray complex, resulting in inaccurate dies in the bucco-lingual dimension. Purpose. This clinical pilot study compared the dimensions of dies fabricated with 3 types of double-arch impressions to dies fabricated with the conventional complete-arch, custom tray method. Material and methods. Thirty-five addition silicone impressions were made of cast metal copings cemented onto natural teeth prepared as complete-crown abutments. Four combinations of tray types and impression material viscosity, were used: ( 1) complete-arch, custom acrylic trays loaded with heavy-bodied material; (2) double-arch, disposable plastic trays loaded with heavy-bodied material; (3) double-arch, disposable plastic trays loaded with putty, material; and (4) double-arch, reusable brass metal trays loaded with heavy-bodied material. Immediately prior to tray insertion, light-bodied impression material v.,as syringed over all copings as a wash. The 4 copings were fabricated from cast gold and simulated metal-ceramic complete-crown thimbles with polished collars and had "projections" on the occlusal surfaces. The impressions were poured in type I-V die-stone. Bucco-lingual and inter-abutment dimensions were measured. The differences between the stone dimensions and cast metal control dimensions A,ere calculated and converted to percent dimensional change. The data were analyzed with 1-way analysis of variance, Student t tests, and Mann-Whitney tests (P<.05). Results. The plastic double-arch tray loaded with heavy-viscosity, addition silicone and a low-viscosity, wash produced the least accurate combination inter- and intra-abutment dimensions. For this protocol, 1.17% mean dimensional change was recorded. This result was significantly different than that obtained for the other 3 impression methods tested. No significant differences were found between the complete-arch method and protocols in which putty was loaded in a plastic or metal tray. Conclusion. Within the limitations of this pilot study, the more rigid tray/impression material combinations more accurately replicated stone dies.