Aim: To compare the accuracy of 3D superimposed models produced by two programs using voxel based superimposition in follow up of orthodontic cases. Patients and methods: The study included 21 non-growing patients performing orthodontic treatment. The gold standard was obtained by tracing each CBCT scan then recording linear measurements between evaluation points. T1 pre-treatment and T2 post-treatment scans were fused forming 3D superimposed model using two software programs; In Vivo (Anatomage, San Jose, California, USA), and Ondemand 3D (Ondemand 3D; Cybermed Co., Seoul, Korea). Accuracy as well as inter-observer reliability between three trained observers were calculated. Results: Cronbach's alpha test showed significant low accuracy of superimposed models performed by In Vivo on canine and premolar readings and those obtained by Ondemand 3D showed significant low accuracy in canine readings. Cronbach's alpha test showed significant moderate inter-observer reliability in canine readings and significant low inter-observer reliability in molar readings performed by InVivo. While inter-observer reliability of on demand 3D was significantly moderate in molar and canine readings. Conclusion: Statistical results showed low accuracy of 3D superimposed models performed by InVivo and Ondemand. In OnDemand showed higher inter observe reliability than InVivo.