This study compared the incidence of deviation along curved canals after preparation with two nickel-titanium (NiTi) rotary systems, Twisted File Adaptive and BT-RaCe. Forty resin training blocks with curved canals were filled with ink and divided into two groups according to the instrumentation technique. Preinstrumentation images were acquired by using a stereomicroscope. The canals were up to an instrument #35/0.04. Postinstrumentation images were captured using the same conditions, and the images were superimposed. The amount of resin removed was measured at 8 different points, beginning at the apical terminus of the canal. Differences in the mesial and distal aspects were measured to evaluate the occurrence of deviation. The Student?s-t test was used for comparison of the intragroup deviation. Intergroup analysis was performed by using one-way ANOVA for each level. For multiple comparisons, the Bonferroni test was used and a cutoff for significance was set at 5%. Intragroup analysis showed that both instrumentation techniques promoted some deviation at all levels. BT-RaCe showed significantly lower deviation at 0 and 4-mm levels than Group Twisted File Adaptive (p<0.05). On the other hand BT-RaCe showed worse performance at level 6. The results of this study demonstrated that none of the NiTi tested systems was able to instrument curved canals simulated in resin blocks without some deviation during the preparation. There is still need for improvement in the instruments manufacturing aiming the better performance of endodontic files in curved root canals.