The aim of this study was to clarify the differences in the phototoxicity of bergamot oil obtained from four different suppliers. Spectral and chemical analyses were performed to identify presence of photoactive compounds in the test samples. The phototoxicity was assessed in vitro by the 3T3 NRU phototoxicity test (PT) and subsequently in a phototoxicity test on reconstructed human skin model (H3D PT). Confirmatory photopatch tests in a group of volunteers were performed using the first non-phototoxic concentration determined in the H3D PT. The spectral and chemical analyses revealed, that two samples of bergamot oil exhibited a potential for photoactivation. These oils were subsequently classified as phototoxic in the 3T3 NRU PT, however, only on the basis of borderline results and depending on the solvent used. H3D PT revealed clear classifications, correlating well with the findings of spectral and chemical analysis. The test was, however, not yet capable of precise prediction of safe, non-phototoxic concentrations. Additional endpoints, e.g. interleukin determination might be employed to increase the sensitivity of the test. Although the study showed the usefulness of the tiered testing strategy, currently, the extrapolation of in vitro results to human situation may be performed only to a limited extent.