This study compared the dimensional accuracy of impressions and dies made from a metal model simulating prepared abutments and having gingival sulci of varying widths. Measurements of the abutments, impressions, and stone dies were made using a travelling microscope, and the number of defects in each impression was recorded. The impressions and dies made from abutments with narrower sulci showed greater distortions. Analysis of variance and the Fisher PLSD post hoc test indicated significant differences between the group having a sulcular width of 0.08 mm and the groups having larger sulcular widths for the impressions and for the dies (P .05). The large coefficient of variation occurring groups having 0.08-, 0.13-, and 0.18-mm sulcular widths demonstrated the difficulty of consistently obtaining good impressions of abutments having such narrow sulcular widths. Between 50% and 90% of abutment impressions having sulcular widths of 0.08 and 0.13 mm had defects.