To assess soil-aquifer treatment of sewage effluent for removal of viruses, studies were conducted at a recharge/recovery site near Tucson, Ariz. Two 13 m2 basins were constructed in coarse sand alluvium, one for secondary- and one for tertiary-treated effluent. Bacterial viruses, MS2 and PRD1, and a chemical tracer, potassium bromide (KBr), were added to effluent applied to these basins. Infiltration rates ranged from 0.2 to 16.8 m/d. Samples of unsaturated flow from depths of 0.30-6.08 m below the basin were taken through porous stainless steel suction-samplers. Bromide and virus results indicated the presence of preferential flow conditions that produced irregular concentration profiles with depth. Virus transport was retarded (R = 1.9) at the beginning of a flooding cycle, but viruses were transported faster than the average water velocity (R = 0.47) when applied after the infiltration rate had declined following 4 days of flooding. Virus specific removal rates (b) during percolation through soil were 2.3-120 times greater than in bottles of effluent or ground water. PRD1 was removed more rapidly during percolation (b = 0.65 h-1) than MS2 (b = 0.23 h-1). Effluent type did not significantly affect b for MS2, but the PRD1 rate was nearly 3 times greater with secondary effluent (1.0 h-1) compared to tertiary effluent (0.35 h-1). Virus removals at the 4.3 m depth ranged from 37 to 99.7%.