Tile drainage and soil amendments have been found to affect losses of nitrate N from agricultural soils. This work was aimed at measuring nitrate N losses in a tile-drained marsh soil from SW Spain under traditional fertilization and irrigation practices, and how these losses were influenced by the application of soil amendments. To this end, a randomised block experiment with three replications was performed during two consecutive growing seasons-2003 to 2004 with cotton and sugar beet, respectivelyinvolving four different amendment treatments: (1) control without amendment, (2) phosphogypsum (PG), (3) manure, and (4) sugar factory refuse lime (SFRL). Flow-weighted (FW) nitrate-N concentrations in drainage water, estimated as the slope of the regression of the instantaneous nitrate-N flow as a function of drain flow rate, was decreased by PG in some drainage events in the 2003 season and in the four last events of the 2004 season when compared with control without amendment. The increased FW nitrate-N concentrations in drainage from SFRL in comparison to control in a drainage event of 2003 season, and in the four last events of 2004, can be explained by the contribution of N present in the amendment. These effects did not account for significant differences in nitrate-N loss among treatments over the whole season in 2003, when they ranged from 19.3 to 24.9 kg N ha -1 , accounting for 6-8% of applied N, nor in 2004, when they ranged from 4 to 6 kg N ha -1 , accounting for 3-4% of applied N. The decrease in mean FW nitrate-N concentration after the third drainage event in 2003 was not the consequence of the depletion of total soil nitrate-N because soil mineral N was increased on average by 205 kg N ha -1 during the season. The high N extractions by sugar beet and the subsequent decrease in total soil nitrate-N can contribute to explain the decrease of mean FW nitrate-N concentrations along the 2004 season. Greater absolute nitrate-N loss in 2003 than in 2004 was explained by the lower efficiency of the furrow irrigation when compared with sprinkler irrigation. Results also revealed that traditional management of N fertilizer was inadequate: rates applied to cotton were excessive, increasing the risk of N losses not only during the cotton season, but also at the beginning of the following season.Keywords Drainage Á Furrow Á Sprinkler Á Nitrate Á Phosphogypsum Á Manure María Dolores Hurtado died in February 2010. The present work was a part of her unfinished Ph. D. Thesis. Her colleagues and friends wish to dedicate this article to her memory.