Background: Drug-induced photosensitivity is difficult to predict and remains a challenge for both the dermatological clinical practice and pharmacovigilance., Purpose: To assess the association between spectroscopic and molecular characteristics and the occurrence of photosensitivity reactions., Methods: For 143 well-known photosensitisers (e.g. tetracyclines, diuretics), we retrieved information on spectroscopic and molecular parameters, including: absorption maximum lambda(max), molar absorption coefficient epsilon, area under the absorption curve (AUC), molecular weight and configuration, hetero and aromatic halogen atoms, lipophilicity (log P) and acid/base status (pKa). In the WHO-ADR database, all reports with suspected adverse drug reactions of the study drugs were selected. We identified all reports on photosensitivity reactions and defined them as cases. All other reports were selected as non-cases. A case-non-case approach was performed to assess the spectroscopic and molecular exposure variables as a factor for photosensitivity reactions. Logistic regression was used to calculate odds ratios (OR) with 95% confidence intervals (CI)., Results: A lambda(max) between 290 and 320 nm (OR 3.74, 95% CI 3.45-4.06), and an epsilon > 20,000 M(-1) cm(-1) (OR 5.49, 95% CI 5.10-5.92) were highly associated with the reporting of photosensitivity reactions. Risk of the photosensitivity reactions was significantly increased among intermediate or high AUCs compared to low AUC. Low molecular weight and aromatic halogen atoms were associated with photosensitivity reactions (OR 2.37, 95% CI 2.07-2.71 resp. OR 3.37, 95% CI 3.15-3.61) as were log p < 1 and pKa < 7., Conclusion: The reporting of photosensitivity reactions to established phototoxic drug classes is strongly influenced by spectroscopic and physicochemical characteristics of individual drugs., ((c) 2009 John Wiley & Sons, Ltd.)