Nitrate leaching from agricultural soils can represent a substantial loss of fertilizer nitrogen (N), but a large variation in losses has been reported. We report N leaching losses under four N fertilizer treatments and two farmer's fields in the Yaqui Valley, Mexico. In these irrigated wheat systems, farmers typically apply 250 kg N as anhydrous ammonia (knifed in) or urea (broadcast), with 75% applied directly before planting and 25% at the time of the first post-planting irrigation. Over two wheat seasons, we compared typical farmer's practices to alternatives that applied less N and more closely timed fertilizer application to plant demand. Field lysimeter measurements and predictions from a water transport simulation model (called NLOSS) were used to determine the amount of N leached over the season. Approximately 5 and 2% of the applied N leached below the root zone with the typical farmer's practice in 1995-96 and 1997-98, respectively. The alternative treatments reduced N leaching losses by 60 to 95% while producing comparable economic returns to the farmers. Leaching losses from the two farmer's fields were substantially higher (about 14 and 26% of the applied N). Our results indicate that the typical farmer's practice leads to relatively high N leaching losses, and that alternative practices synchronizing fertilizer application with crop demand can substantially reduce these losses.