Two identical pilot scale sand filters were operated in parallel to study the effects of preozonation in direct filtration. No coagulants were used. The influent of one filter was ozonated, whereas the influent to the second filter was aerated. Significantly improved reduction in turbidity, particle count, and iron was observed with the filter receiving preozonated water. The head loss development rate was increased as a result of using ozone. In a second set of experiments, the effects of ozone injection point on filter performance were investigated. Two identical filters both receiving preozonated water were operated. With one of the filters, the raw water was ozonated immediately before it entered the filter. The entrance of the preozonatea water to the other filter was delayed by using a detention tank between the ozone contact chamber and the filter. In addition to effluent turbidity and particle count values, zeta potentials of the waters entering the two filters were measured. Head loss development at several locations within each filter bed was also observed and recorded. Slightly better effluent quality (turbidity and particle count) was observed with the filter receiving preozonated water without delay. It was observed that the zeta potential of the prezonated water became more negative with increasing delay time.