The UVA protection delivered by sunscreens is an issue of increasing importance due to the increasing knowledge about UVA-induced skin damage. In Europe there is no officially accepted method available to determine the degree of UVA protection. Therefore, the objective of the present study was to design a protocol combining the merits of an in vitro model, which are simple and reproducible, with aspects known to be relevant from in vivo studies. The principle is: an UV-transparent support to which the test product is applied, a (pre)irradiation and a transmission measurement. Transpore(R) tape (standard support for SPF determinations) was found to be incompatible with many preparations on prolonged contact times. Roughened quartz was adopted as a suitable alternative. Transmission measurements on this support are not reliable with a layer of 2 mg cm(-2) (standard for SPF) due to detection limitations of spectrophotometers, hence a reduced layer of 0.75 mg cm(-2) was adopted. Overall, it is very difficult to apply products in a reproducible thin layer on appropriate substrates. As a consequence, absolute parameters derived from the transmission profile show relatively large dispersion, whereas relative parameters, such as critical wavelength lambda(c)[1] or UVA/UVB ratio are much less sensitive to unavoidable variations in layer thickness. An increase in deviations was observed when the samples were irradiated before measurement. It is crucial to control the output carefully (spectral distribution and even more importantly, irradiance and dose delivered) of the light source. By doing so and also taking into account the previous learning steps, a protocol was drafted and tested in a ringtest (four samples in six laboratories). The results are encouraging and show that if relative parameters (e.g. lambda(c), UVA/UVB ratio) are considered, the intra- as well as interlaboratory reproducibility is clearly better than can be obtained in vivo. In general, we describe a suitable method, which can be considered in any future official discussions about the methodology to determine UVA protection.