The paper presents a hybrid experimental/theoretical approach to study the effect of variation due to process parameter change on the dimensional and geometric accuracy of a part built through ‘fused deposition modeling’ process. A theoretical framework comprising of a primitive based fast geometric model and a comprehensive process parameter-geometric tolerance matrix is presented. This framework is tested with help of an experimental setup based on a geometric variation measurement model for major classes for Geometric Dimensioning & Tolerancing (GD&T) tolerances. Finally, a statistical approach based on Taguchi's design of experiment is utilized to study the effect of commonly used process parameters to establish indicative GD&T tolerance bands for the specific printer class as well as ranking of parameters.