The impact of source water dissolved organic matter (DOM) origin, empty bed contact time (EBCT), temperature, and pretreatment methods on biofiltration performance was evaluated and predictive models based on experimental data were developed. Three DOM source water types, terrestrial, microbial, and treated wastewater (WW) effluent, were utilized. A model was developed to predict biofilter performance for dissolved organic carbon (DOC) removal based on the influent biodegradable DOC (BDOC) fraction, a single active biomass measurement from the top of the filter and the filter EBCT. A biomass distribution model was developed to predict total active biomass throughout the filter based on a single biomass measurement from the top of the filter. The measured BDOC fractions were 21 % for the nonWW impacted source waters, 36 % for the WW effluents and 62 % for the ozonated WW effluents. At an EBCT of 15 min, biofilters removed between 7 and 21 % of the DOC (19 to 50 % for BDOC) depending on the DOM type and use of ozonation. When the EBCT decreased to 5 min DOC removal decreased by 40 % and when increased to 30 min removal increased by 42 %. When the temperature decreased from 22 °C to 6 °C DOC removal was 33 % lower and when increased to 28 °C removal was 42 % higher. ATP values were found to be a function of temperature and DOM origin, as the average ATP values from the WW effluent biofilters were almost double that of the non-WW impacted sources and pre-ozonation of the WW effluent yielded values three times higher. The model was applied to the results of 27 different biofilter runs at three EBCTs yielding one distinct rate constant for the non-WW impacted source waters and one rate constant for the WW effluents. The model was successfully applied to the results of 19 filter runs from the literature and to those from a pilot plant over 6 months of operation.