To achieve higher accuracy in power rating of photovoltaic modules and systems we recently developed an advanced PV performance model that separately addresses the contribution of diffuse and direct solar irradiance. To further improve the accuracy, we upgraded the model in this study to include the influence of solar spectrum on PV module conversion efficiency. We measured solar spectrum in Ljubljana separately for diffuse and direct irradiances under different weather conditions and calculated spectral losses or gains compared to AM1.5 reference spectrum. For diffuse irradiance under clear sky conditions we detected spectral losses from 4.5% during morning hours and up to 23.9% at around noon. During cloudy conditions the average losses were 3.5% and the variation of losses during the day was hardly noticeable. For direct irradiance we detected spectral gains, ranging from 0.8% to 10.3% for air mass values from 1.1 to 5.0, respectively. We modeled obtained results of spectral influence on PV module performance by introducing spectral correction factor separately for direct irradiance with air mass as an input parameter and for diffuse irradiance with the share of diffuse irradiance as an input parameter. The spectral correction factors are applied to measured diffuse and direct irradiance which are then used as an input parameters of the developed PV performance model. We evaluated the increase of accuracy of the model with the spectral upgrade for one month and one year measurements of a silicon PV module at an outdoor monitoring test site in Ljubljana, Slovenia. For the month of June 2014 the normalized root-mean square error of the spectrally upgraded model was 2.3% compared to 2.7% without spectral correction. For the eight snowless months period in 2014 the error of evaluated performance model improved from 3.7% to 3.3% when the spectral correction was applied. (C) 2017 Elsevier Ltd. All rights reserved.