Dialysis and chemical speciation modelling have been used to calculate activities of Fe3+ for a range of UK surface waters of varying chemistry (pH 4.3–8.0; dissolved organic carbon 1.7–40.3 mg l−1) at 283 K. The resulting activities were regressed against pH to give the empirical model: . Predicted Fe3+ activities are consistent with a solid–solution equilibrium with hydrous ferric oxide, consistent with some previous studies on Fe(III) solubility in the laboratory. However, as has also sometimes been observed in the laboratory, the slope of the solubility equation is lower than the theoretical value of 3. The empirical model was used to predict concentrations of Fe in dialysates and ultrafiltrates of globally distributed surface and soil/groundwaters. The predictions were improved greatly by the incorporation of a temperature correction for , consistent with the temperature dependence of previously reported hydrous ferric oxide solubility. The empirical model, incorporating temperature effects, may be used to make generic predictions of the ratio of free and complexed Fe(III) to dissolved organic matter in freshwaters. Comparison of such ratios with observed Fe:dissolved organic matter ratios allows an assessment to be made of the amounts of Fe present as Fe(II) or colloidal Fe(III), where no separate measurements have been made. Electronic supplementary material The online version of this article (doi:10.1007/s10498-008-9040-5) contains supplementary material, which is available to authorized users.