The currently used 8% fragrance mix (FM I) does not identify all patients with a positive history of adverse reactions to fragrances. A new FM II with 6 frequently used chemicals was evaluated in 1701 consecutive patients patch tested in 6 dermatological centres in Europe. FM II was tested in 3 concentrations - 28% FM II contained 5% hydroxyisohexyl 3-cyclohexene carboxaldehyde (Lyral), 2% citral, 5% farnesol, 5% coumarin, 1% citronellol and 10%alpha-hexyl-cinnamic aldehyde; in 14% FM II, the single constituents' concentration was lowered to 50% and in 2.8% FM II to 10%. Each patient was classified regarding a history of adverse reactions to fragrances: certain, probable, questionable, none. Positive reactions to FM I occurred in 6.5% of the patients. Positive reactions to FM II were dose-dependent and increased from 1.3% (2.8% FM II), through 2.9% (14% FM II) to 4.1% (28% FM II). Reactions classified as doubtful or irritant varied considerably between the 6 centres, with a mean value of 7.2% for FM I and means ranging from 1.8% to 10.6% for FM II. 8.7% of the tested patients had a certain fragrance history. Of these, 25.2% were positive to FM I; reactivity to FM II was again dose-dependent and ranged from 8.1% to 17.6% in this subgroup. Comparing 2 groups of history - certain and none - values for sensitivity and specificity were calculated: sensitivity: FM I, 25.2%; 2.8% FM II, 8.1%; 14% FM II, 13.5%; 28% FM II, 17.6%; specificity: FM I, 96.5%; 2.8% FM II, 99.5%; 14% FM II, 98.8%; 28% FM II, 98.1%. 31/70 patients (44.3%) positive to 28% FM II were negative to FM I, with 14% FM II this proportion being 16/50 (32%). In the group of patients with a certain history, a total of 7 patients were found reacting to FM II only. Conversely, in the group of patients without any fragrance history, there were significantly more positive reactions to FM I than to any concentration of FM II. In conclusion, the new FM II detects additional patients sensitive to fragrances missed by FM I; the number of false-positive reactions is lower with FM II than with FM I. Considering sensitivity, specificity and the frequency of doubtful reactions, the medium concentration, 14% FM II, seems to be the most appropriate diagnostic screening tool.