Nowadays, many new window components known as complex fenestration systems (CFS), such as laser-cut panels and prismatic films, are considered in order to improve the overall luminous properties of building spaces : detailed studies of CFS remain necessary however to validate their daylighting performance. Physical and virtual models are commonly used to assess the daylighting performance of more conventional daylighting strategies within buildings. Several recent studies have reported significant errors for both physical and virtual modelling procedures, 10% modelling errors leading in both cases to 15% up to 170 % inaccuracy in modelled daylight factors assessment: no similar error analysis was carried out in a systematic way for daylighting strategies involving CFS use. A side lit office room equipped with double glazing and a CFS (laser-cut panel and prismatic film) was mocked-up for that purpose in a daylighting test module. The office room was reproduced by way of a 1:10 scale physical model placed under a scanning sky simulator, as well as a virtual model built-up by the way of Radiance lighting program. Several model parameters were varied, leading to the evaluation of model inaccuracies through a sensitivity analysis. The most significant factor (internal surface reflectance) is considered in this paper, leading to a first set of modeling guidelines.