A mathematical model of biocide action against microbial biofilm was tested experimentally by measuring the response of Pseudomonas aeruginosa biofilm to various doses of monochloramine. Pure culture biofilm was developed in continuous flow annular reactors for 7 days, then treated with a 2-, 4-, or 8-h dose of 2 or 4 mg L(-1) monochloramine. Some experiments investigated repeated treatment. Disinfection and regrowth of the biofilm were observed by sampling the biofilm for viable and total cell areal densities for up to 100 h following the biocide treatment. A phenomenological mathematical model was fitted to experimental data sets and captured overall trends, but it could not simulate certain experimentally observed features. The model did simulate rapid disinfection followed by steady regrowth. It correctly predicted a much greater decrease in viable than in total cell densities and also correctly captured the shapes of these trajectories. Discrepancies between the model and data included the following: the model predicted faster regrowth than was experimentally observed, the model predicted that a second dose would be more effective than the first dose but the opposite was observed in the experiments, and parameters estimated by fitting one dose concentration could not be used to predict the results of a different dose concentration or a second dose. Discrepancies between model and the experiment were hypothesized to be due to an adaptive stress response by the bacteria, a process not included in the model. A practical implication of this work is that it is more effective to deliver monochloramine in a short concentrated dose as opposed to a longer dose of lower concentration. (c) 1997 John Wiley & Sons, Inc. Biotechnol Bioeng 56: 201-209, 1997.