The electron spin resonance (ESR)-based photosafety test (ESR-PT) is a non-animal prediction test for photosafety evaluations that can be used even for hydrophobic chemicals; the method is based on the detection of singlet oxygen generation using ESR spectroscopy and showing high accuracy for compounds with known photosafety information. During the process of extending the application data for ESR-PT, we found three false-negative chemicals: bithionol, fenticlor and cilnidipine. These chemicals did not show the characteristic triplet signal of 4-hydroxy-2,2,6,6-tetramethyl-piperidine-1-oxyl (4-hydroxy-TEMPO), which is used as a classifier for positive or negative chemicals; instead, bithionol and fenticlor induced an apparent single peak signal with a g-value of 2.0048, while cilnidipine produced a small, fragmented signal. Bithionol and fenticlor reportedly induce free radicals, and positive phototoxic or photoallergic evidence have been reported. Although the small, fragmented signal observed for cilnidipine was confirmed to be identical to that of a phenylnitroxy radical by the computer simulation, the significance of this chemical for photosafety considerations may be low because cilnidipine has quite a low incidence of phototoxic or photoallergic reactions in humans. Accordingly, the current ESR-PT protocol should be improved to detect free radical photoproducts generated from chemicals such as bithionol and fenticlor, thereby helping to reduce false negatives in ESR-PT.