Stereoscopic video endoscopes are widely used for remote visual inspection and precise three-dimensional (3D) measurements in industrial and biomedical applications. The reconstruction of 3D points from the corresponding image points requires a geometrical calibration procedure, the accuracy of which affects the measurement uncertainty. We propose to perform an optimal choice of the calibration technique and the calibration target parameters using a computer simulation at the design stage. The effectiveness of this approach is demonstrated via the design of a self-developed miniature prism-based stereoscopic system. We simulated acquisition of calibration and measurement data using optical design software. The conventional calibration technique, requiring many positions of the flat target with arbitrary displacements and rotations, was compared with another one that uses the translation stage to provide pure translation of the target. We analyzed the impact of the translation uncertainty, the number of positions, the number of targets, and the uncertainty of image point coordinates on the uncertainty of calibration parameters and 3D measurements. We have shown that the second technique could provide acceptable measurement accuracy using only two images. The results of the computer simulation were confirmed experimentally using the prototype of the stereoscopic endoscope. The proposed approach may be used to optimize calibration techniques and reduce the cost of calibration equipment for various stereoscopic measurement systems.