A detailed study has been conducted to assess the effect of various artificial weathering cycles on the degradation of three formulations of window-grade UPVC. The results were compared with samples which had been exposed outdoors in the north-east of England for periods of up to 16 years. Microhardness, loss of calcium carbonate filler, intensity of the C-O stretching band, colour and gloss were used as measures of degradation. The samples tested in the laboratory exhibited the characteristic surface degradation and partial recovery predicted for these materials. Cyclic variation of certain properties was exhibited for long laboratory exposures. In contrast, the naturally exposed material gave no definitive indication of a recovery process. The various measures of degradation employed gave reasonably consistent indications of degradation, particularly with regard to the onset of partial recovery. The complexity of the degradation process for these materials inherently limits the establishment of quantitative rate equations to predict the durability of naturally exposed materials. Nevertheless, four of the five exposure cycles gave consistent ranking of the different materials which compared well with the ranking deduced from natural exposure. An optimum artificial weathering cycle was derived. It is recommended that laboratory tests should involve testing of a reference sample in parallel for the purposes of quality control and interlaboratory comparison. By choosing a material with established outdoor performance, a basis for prediction of relative service life is also provided.