Objective. To investigate the accuracy of dies obtained from single-step and 2-step double-mix impressions. Material and Methods. Impressions (n = 10) of a stainless steel die simulating a complete crown preparation were performed using a polyether (Impregum Soft Heavy and Light body) and a vinyl polysiloxane (Perfectim Blue Velvet and Flexi-Velvet) in two consistencies, in one or two (without relief) steps. Accuracy of the stone dies was accessed at a measuring microscope, using a metallic crown with perfect fit to the reference crown preparation. Data were submitted to 2-way ANOVA and Tukey test (a = 0.05). Results. The single-step technique resulted in slightly larger dies, while the 2-step technique without relief produced significantly smaller dies, when compared to the original stainless steel die. Stone dies obtained from 2-step polyether impressions were significantly smaller when compared to dies obtained from 2-step vinyl polysiloxane impressions (Impregum 2-step: -290.94 ± 71.64¿µm; Perfectim 2-step: -201.86 ± 28.58¿µm). No significant differences were observed in dies obtained from either polyether or vinyl polysiloxane with the single-step technique (Impregum single-step: 63.52 ± 16.60¿µm; Perfectim single-step: 79.40 ± 14.11¿µm). Conclusion. Higher discrepancies were detected for the 2-step impression technique without relief for the investigated materials.