Waterfowl habitat conservation strategies in the Mississippi Alluvial Valley (MAV) and several other wintering areas assume carrying capacity is limited by available food, and increasing food resources is an effective conservation goal. Because existing research on winter food abundance and depletion is insufficient to test this hypothesis, we used harvested rice fields as model foraging habitats to determine if waste rice seed is depleted before spring migration. We sampled rice fields (n  =  39 [winter 2000–2001], n  =  69 [2001–2002]) to estimate seed mass when waterfowl arrived in late autumn and departed in late winter. We also placed exclosures in subsets of fields in autumn (n  =  8 [2000–2001], n  =  20 [2001–2002]) and compared seed mass inside and outside exclosures in late winter to estimate rice depletion attributable to waterfowl and other processes. Finally, we used an experiment to determine if the extent of rice depletion differed among fields of varying initial abundance and if the seed mass at which waterfowl ceased foraging or abandoned fields differed from a hypothesized giving-up value of 50 kg/ha. Mean seed mass was greater in late autumn 2000 than 2001 (127.0 vs. 83.9 kg/ha; P  =  0.018) but decreased more during winter 2000–2001 than 2001–2002 (91.3 vs. 55.7 kg/ha) and did not differ at the end of winter (35.8 vs. 28.3 kg/ha; P  =  0.651). Assuming equal loss to deterioration inside and outside exclosures, we estimated waterfowl consumed 61.3 kg/ha (48.3%) of rice present in late autumn 2000 and 21.1 kg/ha (25.1%) in 2001. When we manipulated late-autumn rice abundance, mean giving-up mass of rice seed was similar among treatments (48.7 kg/ha; P  =  0.205) and did not differ from 50 kg/ha (P  =  0.726). We integrated results by constructing scenarios in which waterfowl consumed rice at different times in winter, consumption and deterioration were competing risks, and consumption occurred only above 50 kg/ha. Results indicated waterfowl likely consumed available rice soon after fields were flooded and the amount consumed exceeded our empirical estimates but was ≤48% (winters pooled) of rice initially present. We suggest 1) using 50 kg/ha as a threshold below which profitability limits waterfowl feeding in MAV rice fields; 2) reducing the current estimate (130 kg/ha) of rice consumed in harvested fields to 47.2 kg/ha; and 3) increasing available rice by increasing total area of fields managed, altering management practices (e.g., staggered flooding), and exploring the potential for producing second or ratoon rice crops for waterfowl.