The long-term removal of 34 trace organic micropollutants (<1 μg L(-1)) was evaluated and modeled in drinking water biological filters with sand media from a full-scale plant. The micropollutants included pesticides, pharmaceuticals, and personal care products, some of which are endocrine disrupting chemicals, and represent a wide range of uses, chemical structures, adsorbabilities, and biodegradabilities. Micropollutant removal ranged from no measurable removal (<15%) for 13 compounds to removal below the detection limit and followed one of four trends over the one year study period: steady state removal throughout, increasing removal to steady state (acclimation), decreasing removal, or no removal (recalcitrant). Removals for all 19 nonrecalcitrant compounds followed first-order kinetics when at steady state with increased removal at longer empty bed contact times (EBCT). Rate constants were calculated, 0.02-0.37 min(-1), and used in a pseudo-first-order rate model with the EBCT to predict removals in laboratory biofilters at a different EBCT and influent conditions. Drinking water biofiltration has the potential to be an effective process for the control of many trace organic contaminants and a pseudo-first-order model can serve as an appropriate method for approximating performance.