Synopsis The sun protection factor (SPF) of two products, one with an expected SPF of 4 and another with an expected SPF of 15 were determined, using two solar simulators: Multiport Solar UV Simulator (xenon, Solar Light, Philadelphia, PA, USA), and Supersun 5000 (metal halide, Mutzhas, Munich, Germany). The mean SPFs using the Multiport were: 4.8 for the low SPF product and 19.4 for the high SPF one. The results using the Supersun were lower: 2.6 for the low SPF product and 7.2 for the high SPF one. Relative emission spectra of the two sources were recorded using a fluorescence spectrophotometer in bioluminescence mode. Efficacy spectra were calculated and compared with the corresponding spectrum of natural sunlight. It was evident that the spectral power of the xenon source is too high in the UVB, leading to overestimation of SPFs, whereas the Supersun irradiator emits too much in the UVA, resulting in too low SPFs. Heat effects and photodegradation of UV filters are discussed as further possible reasons for the discrepancies between the experimentally determined SPFs. Our results confirm a recent publication about theoretical SPFs, calculated with emission spectra of a xenon source and spectra of the sun at different elevation angles, where the authors provide evidence that in natural sunlight the contribution of UVA to total UV radiation is twice as high as in a xenon source. This may contribute to an understanding of why sunscreens tested according to the FDA method (xenon sources) often yield higher SPFs than those obtained from European testing procedures.