High spatial and temporal variability of nitrogen application via livestock urine on grazed pasture make meaningful measurement of leaching of nitrate very difficult. The present study was aimed at estimating the mean annual nitrogen loadings and concentrations leached to an underlying aquifer beneath unfertilized irrigated and non-irrigated paddocks of a dairy farm at OB Flat 10 km south of the city of Mount Gambier (37°50′S, 140°40′E), South Australia. Urine was collected in the milking shed and applied to 1 m deep monolith lysimeters in three separate seasons in 1992. Some lysimeters received two applications. Drainage and N fluxes and concentrations were recorded 1 year before urine application and for 3 years following urine applications. A negative binomial function was used to specify the appropriate proportions of pasture subject to 0, 1 or 2 loadings in a year, and the lysimeter drainage measurements were then scaled across seasons and paddock area to estimate mean annual leaching under the whole of each paddock. It was found that the timing of urine application had a profound effect on the leaching of nitrogen. Results indicated that the nitrate flux leaching from the root zone in the irrigated paddock was 26–33 kg N/ha/year at 13–17 mg N/litre and in the non-irrigated paddock, 10–13 kg N/ha/year at 9–11 mg N/litre. These estimates are not inconsistent with nitrate concentrations measured in groundwater but cannot be independently verified. The current transition from non-irrigated to irrigated pastures is likely to double nitrogen loads to the aquifer and cause nitrate concentrations in groundwater to exceed drinking water guidelines, unless alternative pasture management strategies are developed.