This article covers a comparative analysis of the potentials and constraints of different heat sources (air, geo-thermal boreholes, lake, river, groundwater and solar thermal) exploited by HP systems, implemented in various types of multifamily buildings (MFB) - new, retrofitted and non-retrofitted - which correspond to real case studies situated in Geneva. After characterizing the various heat sources and building demands, as well as presenting the numerical model and adopted sizing values, we study the intrinsic potential of the various HP heat sources and show that the HP seasonal performance factor (SPF) is directly correlated to the heat source temperature. In a further step we consider complementary PV production for the HP system, taking into account the available roof area and daily profile match. For buildings with a combined space heating and domestic hot water heat demand up to 80 kWh/m(2), which correspond to current best case buildings (10% of the existing MFB stock in Geneva), combined HP & PV systems should lead to an annual purchased electricity inferior to 15 kWh/ m(2) (with a factor 2 between best and worst heat sources), with an associated daily peak load up to 150 Wh/m(2)/ day. For a demand below 130 kWh/m(2) (which is the case of 75% of the existing MFB stock of the Canton), the various combinations of HP & PV systems mainly result in a purchased electricity below 45 kWh/m(2). The daily peak load reaches up to 500 Wh/m(2)/day, or eventually higher in the case of high-rise buildings. Aside from the final purchased electricity, the annual electricity injected into the grid is in the order of 15-20 kWh/m(2) for low-rise buildings, and half that much for high-rise buildings (except for solar HP systems, for which the reduced available roof area for PV leads to significantly lower values). Lastly, SPF alone is not a sufficient indicator for the characterization of the HP system performance, since it doesn't reflect the absolute value of the electricity demand, which primarily depends on the building heat demand. Furthermore, both SPF and annual electricity demand are limited to annual balance considerations. As a complement, an indication of the peak electricity load gives valuable indications of the potential stress on the grid.