Nitrogen (N) inputs have been found to exert strong influence on leaf stoichiometry in natural ecosystems, but there are few studies investigating the effects of N in agroecosystems. Using a 5-yr fertilization experiment in rice fields, we examined the effects of N inputs on leaf stoichiometry of one crop, rice (Oryza sativa L.), and its four common weeds, barnyardgrass [Echinochloa crus-galli (L.) P. Beauv.], Monochoria korsakowii Regel and Mack, alligatorweed [Alternanthera philoxeroides (Mart.) Griseb.], and Japanese mazus [Mazus pumilus (Burm. f.) Steenis], and further evaluated whether and how straw return mediates these effects. We found that rice and weed leaf nitrogen:phosphorus:potassium (N:P:K) stoichiometry exhibited divergent responses to N fertilizer. Weed leaf N:P:K stoichiometry was not sensitive to low (120 kg N ha^-1) and regular (240 kg N ha^-1) N inputs, but rice plants were, with significantly increased leaf N concentration and N:P and N:K ratios. The opposite trend was found for high N inputs (360 kg N ha^-1). Rice leaf N concentration [N] did not increase further, and N:P ratios even decreased, whereas E. crus-galli and M. korsakowii had significantly increased [N] and N-related stoichiometry. We also found that the positive effects of regular N inputs on rice leaf N:P and N:K ratios were significantly dampened by straw return, but the positive effects on N:P ratios in M. pumilus leaves were enhanced by straw return. Compared with weeds, rice leaves contained low elemental concentrations across fertilization levels at grain-filling stages. These results indicate that rice has a lower N requirement than weeds at grain-filling stages, and the N supply should be managed at a relative low level to reduce the nutrient acquisition and competitive abilities of weeds. From a stoichiometric perspective, this study highlights the importance of N management in combination with straw return in controlling weeds and increasing the nutrient-use efficiency of crops.