This paper presents optimization results for a two-phase, modular transverse flux switched reluctance motor (TFSRM) with an outer rotor. In particular, the main disadvantage of the considered motor structure, that is the zero starting torque in some rotor positions, is eliminated by construction optimization. A numerical model of the motor developed in the Flux3D program is coupled with a Matlab-based evolutionary algorithm for optimization of construction parameters of the magnetic circuit. The elaboratem algorithm is also connected with a database to limit the computation costs. Three objective functions are taken into account for the motor integral parameter improvement. The fundamental role of a type of an optimization criterion function is comparatively analyzed and a new effective criterion function is introduced.