Abstract In this paper, two probabilistic reliability indices are presented to define the effect of clouds on different types of Building Integrated Photovoltaic (BIPV) systems. Existing indices do not match the main features of photovoltaic (PV) sources, such as variability, uncertainty and dependency on weather conditions. In addition, they are time indicators that describe the percentage of total failure time per year without any indication about power or energy mismatch. By using the available Geographic Information System solar-irradiation database, the proposed indices consider a similar pattern of expected daily solar irradiation as a model of PV systems. Two different models are studied for diverse building loads: an energy model for flexible loads and a constant-power model for critical loads. A comparative study is implemented for six different locations: Cairo, London, Berlin, Beijing, Madrid and Riyadh. Moreover, three types of BIPV systems are studied: fixed, double-axis-tracking and concentrated PV systems. The presented results show the effects of clouds, PV type and locations on the system performance.