Flow rate, electron donor addition, and biomass control were evaluated in order to optimize perchlorate (ClO4-) removal from drinking water using biologically active carbon (BAC) filtration. Influent dissolved oxygen (DO) was lowered from ambient conditions to approximately 2.5 mg/L for all experiments using a nitrogen sparge. When influent nitrate concentration was 0-2.0 mg/L, 1.6-2.8 mg/L as carbon of acetate or ethanol was required to achieve and sustain the complete removal of 50 microg/L perchlorate in a BAC filter. Most or all of the exogenous acetate and ethanol was removed during biofiltration. When a 72-h electron donor feed failure was simulated, a maximum perchlorate breakthrough of 18 microg/L was observed and, once electron donor was reapplied, 9 days were required to reestablish complete perchlorate removal. During a 24-h electron donor feed failure simulation, the maximum effluent perchlorate concentration detected was 6.7 microg/L. Within 24 h of reactivating the electron donor, the filter regained its capacity to consistently remove 50 microg/L perchlorate to below detection. Although biomass growth diminished the filter's ability to consistently remove perchlorate, a cleaning procedure immediately restored stable, complete perchlorate removal. This cleaning procedure was required approximately every 50 days (4800 bed volumes) when influent DO concentration was 2.5 mg/L. Empty-bed contact time (EBCT) experiments showed that 80% perchlorate removal was achieved using a 5-min EBCT, and complete perchlorate removal was observed for an EBCT of 9 min. It was also demonstrated that BAC filtration consistently removed perchlorate to below detection for influent perchlorate concentrations ranging from 10 to 300 microg/L, influent sulfate concentrations between 0 and 220 mg/L, influent pH values of 6.5-9.0, and operating temperatures of 5-22 degrees C.