Despite the enormous increase in sunscreen use, allergic contact (AC) and photoallergic (PA) reactions to ultraviolet (UV) filters are considered rare. To analyse the data from 2715 patients who underwent photopatch testing at St John's Institute of Dermatology during the period 1983-98. A retrospective analysis of all positive photopatch test episodes was undertaken with the results retrieved from the environmental dermatology database and further verified with the original archived patch test documentation for each individual patient. In 111 patients with positive reactions (4.1%), there were 155 AC or PA reactions to allergens in the photopatch test series. Eighty PA reactions were observed in 62 (2.3%) patients (32 men and 30 women, age range 28-75 years), with UV filters accounting for 52 positive reactions (65%), drugs 16 (20%), musk ambrette 11 (14%) and the antiseptic trichlorocarbanilide one (1%). The most common UV filter photoallergen was benzophenone-3 with 14 positive results, followed by benzophenone-10 (n = 9), isopropyl dibenzoylmethane (n = 6), p-aminobenzoic acid (PABA) (n = 5), octyl dimethyl PABA (n = 5), butyl methoxydibenzoylmethane (n = 4), isoamyl methoxycinnamate (n = 2), ethyl methoxycinnamate (n = 2), octyl methoxycinnamate (n = 2), amyl dimethyl PABA (n = 2) and phenylbenzimidazole sulphonic acid (n = 1). A similar number of AC reactions to UV filters was detected in this study. Thus 49 patients (1.8%) had a total of 75 reactions: 51 due to UV filters and 24 as a result of exposure to fragrances and therapeutic agents. Benzophenone-10 accounted for 13 AC reactions and benzophenone-3 for eight reactions. Twenty-two patients had a PA reaction alone, whereas 19 patients had chronic actinic dermatitis and 15 patients polymorphic light eruption (PLE) in addition. Thus, 34 of the 62 patients (55%) had a preceding underlying photodermatosis. These results show a low yield of positive photopatch tests. Thus, despite the large increase in the use of UV filters over the last decade, the development of PA reactions remains rare. Furthermore, most of the common UV filter photoallergens identified in this study, including PABA, amyl dimethyl PABA and benzophenone-10, are now rarely used in sunscreen manufacture, while isopropyl dibenzoylmethane was voluntarily removed from the market in 1993. Currently, benzophenone-3 is the commonest contact photoallergen still in widespread use. In contrast, the UVB filter octyl methoxycinnamate, used in a number of sunscreens, produced only two positive PA reactions in 12 years of testing. Nevertheless, although these reactions are extremely rare, patients with photodermatoses such as PLE and chronic actinic dermatitis do represent a group of patients at increased risk of developing photoallergy. Further photopatch test series should be regularly reviewed and updated, as the relevance of individual photoallergens changes over time. Currently, there is no evidence that PA reactions represent a common clinical problem.