Removal of fecal coliforms and fecal streptococci was monitored over a period of 13 months in 14 buried pilot scale filters, treating septic tank effluent. The effects of grain size, hydraulic dosing rate and distribution method were investigated. Two different natural sands (sorted sand and unsorted sand) and three different types of light weight aggregates (LWA 0-4 mm, LWA 2-4 mm and crushed LWA 0-3 mm) were used. Intermittent dosing rates from 20 to 80 mm/day in 12 doses per day were applied to the filters by uniform pressure distribution or point application by gravity dosing. Removal of fecal coliforms was more than three orders of magnitude higher in the media with the finest grain sizes (unsorted sand) as compared to the coarsest media (LWA 0-4 mm and LWA 2-4 mm) operated under same conditions. Fecal streptococci were determined only in effluent from filters with LWA 0-4 mm and LWA 2-4 mm. Higher removal of fecal coliforms was observed in pressure dosed filters compared to gravity dosed filters. A lower removal was observed by increasing the hydraulic dosing rate. Minimum retention time was found to be a key parameter for predicting removal of bacteria in unsaturated, aerobic filters. At minimum retention times lower than about 50 h, there was a correlation of 0.96 between retention time and removal of fecal coliforms. Retention times longer than 50 h gave almost complete removal of fecal coliforms.