Tandem solar cells are the best approach to maximize the light harvesting and adjust the overall absorption of the cell to the solar irradiance spectrum. Usually, the front and back subcells are connected in series in two-terminal device (2T) designs which require a current matching between both subcells in order to avoid potential losses. Alternatively, they can also be connected in parallel giving rise to a three terminal connection (3T). In principle, both designs have their assets and drawbacks in terms of device performance, design and materials' characterization. In this letter, we theoretically and experimentally confront both designs with each other (2T and 3T). Theoretical estimations show a maximum PCE of 15% for 2T and about 13% for 3T structures with ideal bandgaps for the front and back cell. However, 3T tandem devices can yield higher efficiencies than 2T for some specific material combinations whose theoretical values are between 10% and 12%. Therefore, other aspects related to the fabrication feasibility are studied in order to analyze the most convenient approach for module development. The need of a conducting interlayer restricts the width of the cell and causes a 3% reduction in the geometrical fill factor of the module in comparison to the 2T approach. The R2R processing of modules with 3T cells would also require an additional laser step. Finally, a couple of existing material combinations have been experimentally implemented into 2T and 3T tandem devices. The limitation imposed by their specific and non-ideal bandgaps restricts the efficiency to around 7%, considerably below the ideal case. (C) 2014 Elsevier B.V. All rights reserved.