Little data are available on P losses in runoff from land under conservation tillage (CT) where the surface-applied fertilizer variable has been eliminated. Thus, simulated rainfall was used to evaluate the comparative effects of four tillage systems on the losses of total P, dissolved molybdate-reactive P (DMRP) and algal-available P (AAP) where fertilizer was subsurface banded at planting. Tillage treatments included conventional (CN) and three CT systems: chisel plow (CH), till-plant (TP) and no-till (NT). Above-ground portions of corn (Zea mays L.) plants were removed prior to simulation. Trials were conducted over a 4-yr period, with individual trials being performed in June and July, September, or October of various years. The NT, CH, and TP treatments reduced total P losses by an average of 81, 70, and 59%, respectively, relative to CN. Concentrations and losses of total P among tillage treatments followed those for sediment concentrations and losses. Concentrations of DMRP were, in most cases, lowest for CN, although differences among treatments were generally small and not significant. With the exception of the first simulated rainfall trial (September 1980), no significant correlation was observed between DMRP concentrations and residue cover. Losses of DMRP for the CT treatments were similar to, or significantly lower than those for CN. Differences in AAP concentrations varied among treatments and sampling periods, although concentrations were usually lowest for NT. The relative percentage of AAP concentrations to total P concentrations increased over the course of this study. This increase was 6, 8, 26% higher for CH, TP, and NT, respectively, relative to CN. Relative to CN, the NT, CH, and TP treatments reduced AAP losses by an average of 63, 58, and 27%, respectively. These results indicate that CT systems can effectively reduce AAP losses in runoff relative to CN, especially at times when high sediment concentrations and losses occur from conventionally tilled land.