We tested the hypothesis that decomposition in flood-inundated patches of riparian tree leaf litter results in higher plant-available nitrogen in underlying, nutrient-poor alluvium. We used leafpacks (n = 56) containing cottonwood (Populus deltoides ssp. wislizenii) leaf litter to mimic natural accumulations of leaves in an experiment conducted on the Yampa River floodplain in semi-arid northwestern Colorado, USA. One-half of the leafpacks were set on the sandy alluvial surface, and one-half were buried 5 cm below the surface. The presence of NO₃⁻ and NH₄⁺ presumed to result from a leafpack's submergence during the predictable spring flood pulse was assessed using an ion-exchange resin bag (IER) placed beneath each leafpack and at control locations. Leafpacks and IERs were collected one week after flood peak (71 days total exposure) at half the stations; the remainder were collected three weeks later (93 days exposure). A multi-peaked spring flood with above-average maximum discharge inundated leafpacks for total time periods ranging from 133 to 577 hours. Litter lost from 43 to 68 percent of its initial organic matter (OM) content. Organic matter loss increased with total time inundated and total time of exposure on the floodplain. Burial retarded OM loss if the total time inundated was relatively long, and substrate texture (sand vs. silt) affected OM loss in a complex manner through interactions with total time inundated and total time of exposure. No pulse of N attributable to leaf breakdown was detected in the IERs, and leafpack litter showed no net change in the mass of nitrogen present. Patterns of leafpack and IER nitrogen levels suggested that litter removed N from floodwater and thereby reduced N availability in underlying sediment. Immobilization of floodwater-N by litter and N mineralization outside the flood period may be important components of N flux in semi-arid and arid floodplain environments.