There is a strong correlation between the number of animals per area unit and Phosphorus (P) contribution to waters (Issermann, 1990). In Occidental Europe, between 27 and 38% of the P reaching water sources comes from agriculture. In Southern Chile, 60% of the dairy production and 45% of the beef production is based on grazing of permanent pastures in volcanic soils, but no runoff studies looking at P transfer from these systems to surface waters have been carried out. The objective of this experiment was to quantify P losses in runoff in beef production systems as affected by the field slope on a volcanic soil. The experiment was carried out between April 2004 and November 2005, on an andisol of the Osorno soil series (40ºS, 73ºW). Two slopes were tested, 3 and 12%. Grazing was carried out with Holstein-Friesian steers (212 kg initial live weight) under rotational grazing (3.5 steers ha-1, in both treatments), in a permanent pasture fertilized with 75 kg N (autumn and spring) and 30 kg P (spring). In each treatment, three surface lysimeters (5x5m) were established (Scholefield & Stone, 1995), and runoff collected three times per week at the ground level. All runoff samples were stored at 4ºC until individual analysis for reactive (RP) and total P (TP). Organic P (OP) was estimated as the differences between TP and RP. Phosphorus losses were calculated as the product of drainage and P concentration in the samples. Analysis Of Variance (ANOVA) was used to compare P concentrations and losses between treatments. Results of the first drainage season showed that the total rainfall for the experimental period was 868 mm and total drainage during the same period was 634 mm in both treatments. Runoff represented 1% only of total drainage. No difference in the contribution of surface runoff to the total drainage was found between treatments (P>0.05). During the first drainage season the field slope increased the average RP and OP by 250% each, from 1.0 to 3.5 mg L-1 for RP and from 0.4 to 1.4 mg L-1 for OP (P<0.05). On average, the proportion of RP in the total amount of P lost in runoff was not different between treatments (75%). In the present experiment, the field slope affected P losses in runoff, so that a three times increased in the field slope increased P losses in runoff eight times. Overall P losses were 0.008 and 0.001 kg P ha-1, for the 12% and 3% slope treatments, respectively (P<0.05). The low amount of P lost from grazing systems in volcanic soils are low, probably because of a combination of both, the high water holding capacity of these soils, which prevents runoff, and the high P fixation capacity of volcanic soils, so that no RP is available to be lost by runoff. Samples from the second drainage season are currently under evaluation, so that the complete set of data will provide a better understanding of the processes and magnitudes of P losses under Chilean conditions. Acknowledgment: This research was funded by FONDECYT, Grant 1040104.