The mineralization of nitrogen (N) and especially the regeneration of ammonium are critical processes performed by bacteria in aquatic ecosystems. Quantifying these processes is complicated because bacteria simultaneously consume and produce ammonium. Here we use experimental data on the effects of the molecular composition of the supplied substrates, combined with a classical stoichiometric model of ammonium regeneration, to demonstrate how the quantification of these processes can be improved. We manipulated a batch culture experiment with an isolated bacterial community by adding three different types of N substrates: dissolved inorganic nitrogen (DIN, nitrate), dissolved organic nitrogen (DON, amino acid) and a mixture of DIN and DON. With such experiment setup , the ammonium regeneration per se could be easily tracked without using complicated methods (e.g. isotope dilution). We compared the experimental data with the predictions of Goldman et al.' model (Limnol Oceanogr 32:1239-1252, 1987) as well as with a revised version, using the measured consumption carbon:nitrogen ratio (C:N ratio), rather than an estimated consumption ratio. We found that, for all substrates, and in particular, mixed substrates where C and N are partially dissociated between different molecules, estimates of ammonium regeneration rates can be improved by measuring the actual consumption C: N ratio.