Architects need support for the evaluation of the daylighting performance of their buildings, As experimental tools, sky simulators, used in conjunction with scale models, allow a concrete and intuitive approach to the problem. Yet conventional artificial skies and their measurement technology are of limited use, because they are usually not able to evaluate building performance under a non-standard sky luminance distribution (CIE sky models). The Partial Daylight Factor (PDF) method presented in this paper addresses this problem. It is applied to a scanning sky simulator in conjunction with a methodology of experimental research (experimental plans of Hadamard). Scale model experimental simulations carried out with the scanning sky simulator and the PDF method offer an evaluation of the daylighting performance of a building for any luminous configuration of the sky vault. Coupled to a climatic database, they allow an evaluation of daylighting performance indicators such as daylight sufficiency and average or cumulated illuminance for all types of overcast, intermediate and clear skies, including CIE standard models. Moreover, new concepts, such as the Partial Illuminance Factor (PIF), allow a better knowledge and comprehension of the daylighting strategy. (C) 2002 Elsevier Science Ltd. All rights reserved.