IntroductionPlaster models are tools of fundamental relevance for diagnose and orthodontic treatment planning [1][2][3]. Dental models are employed for evaluation of sagital, vertical and transversal dimensions and also for crowding or spacing measurements. However, plaster models present some disadvantages, such as huge space for storage, risk of loss or breakage and possibility of mold proliferation over long term [2].In this context, it has been proposed the use of digital models as a tool for diagnose and planning for daily orthodontic practice [1,4]. Digital models do not require physical space for storage, allow easy information exchange among professionals [2] and enable simple and practical digital diagnose setup making [5].Digital model obtainment can be produced by laser scanning of plaster models [6], impressions [7] or intra oral direct scanning [8]. Digital models extracted from computed tomography has also been reported [7].Majorly, the published studies have focused on evaluating the accuracy of digital models obtained from plaster models scanning [6,9,10] and few studies have investigated de precision of digital models produced through dental impression scanning [12,13]. Therefore, diagnose and treatment planning can be compromised when using tools that are not properly calibrated. The present study thus aimed at evaluating the accuracy of alginate impression-obtained virtual models. AbstractIntroduction: Digital models have been proposed as substitutes for plaster models. Dental arch impression scanning is a rapid and practical approach for digital model obtainment; however, few studies have accessed the accuracy of the method. This study verified the accuracy of virtual measurements obtained with the scanner Ortho Insight 3D, version 5.0 (Motionview Software, LLC, Chattanooga, Tennessee, USA). Materials And Methods: Total of 26 plaster models belonging to the Bahia Federal University Orthodontic postgraduation program were divided into three groups: G1 (plaster models); G2 (alginate impressions scan) and G3 (plaster models scan). Virtual measurements were compared to the manual by evaluating upper intercanine and intermolar distances and the antero posterior distance between upper left canine and upper left molar. Paired Student t and Lin agreement (5% of significance level) were employed for the statistical analysis. Results: Of the evaluated measures, the mean difference ranged from 0.48mm to 0.55mm when compared G2 and G1. The mean difference ranged from 0.6mm to 0.65mm between G3 and G1 groups. Comparing the groups G2 and G3 to G1, it was found statistically significant differences between all variables (p <0.05). There was a substantial agreement (ρc between 0.95-0.99) and almost perfect (ρc> 0.99) for all parameters evaluated. Conclusion: Despite having existed significant differences between the virtual and physical approaches, substantial and almost perfect agreement between them allow us to suggest that there is accuracy of virtual images obtained from scanned impressions in alginate ...