The implications of adopting alternative seeding methods for rice and wheat establishment were examined at three geographically separate sites in the rice–wheat system of the Indo-Gangetic plains, across northern India. Rice yields in cultivated plots, established by either wet or dry seeding methods, were evaluated in comparison to yields from zero-tillage plots and under conventional transplanting methods. In the same trials, the effects of crop establishment methods in wheat were assessed both on wheat yields and rice yields. Rice crop establishment methods markedly influenced the emerging weed flora and attainable yields were measured in relation to intensity of weed management. Over four years, average rice grain yields in the absence of weed competition were greatest (6.56 t ha −1) under wet seeding (sowing pre-germinated rice seed on puddled soil), and similar to those from transplanted rice (6.17 t ha −1) into puddled soil, and dry seeded rice after dry soil tillage (6.15 t ha −1). Lowest yields were observed from dry seeded rice sown without tillage (5.44 t ha −1). Rice yield losses due to uncontrolled weed growth were least in transplanted rice (12%) but otherwise large (c. 85%) where rice had been sown to dry cultivated fields or to puddled soil, rising to 98% in dry seeded rice sown without soil tillage. Weed competition reduced multiple rice yield components, and weed biomass in wet seeded rice was six-fold greater that in rice transplanted into puddled soil and twice as much again in dry seeded rice sown either after dry tillage or without tillage. Wheat grain yields were significantly higher from crops sown into tilled soil (3.89 t ha −1) than those sown without tillage (3.51 t ha −1), and also were elevated (5% on average) where the soil had been dry cultivated in preparation for the previous rice crops rather than puddled. The method of wheat cultivation did not influence rice yield. Soil infiltration rates in the wheat season were least where the land had been puddled for rice (1.52 mm h −1), and greater where the soil had been dry-tilled (2.63 mm h −1) and greatest after zero-tillage (3.54 mm h −1). These studies demonstrated at research managed sites across a wide geographic area, and on farmers’ fields, that yields of dry seeded rice sown after dry cultivation of soil were broadly comparable with those of transplanted rice, providing weed competition was absent. These results support the proposition that direct seeding of rice could provide an alternative to the conventional practice of transplanting, and help address rising costs and threats to sustainability in the rice–wheat rotation. Further, analysis of patterns of long-term rainfall data indicated that farmers reliant on monsoon rainfall could prepare fields for dry direct seeded rice some 30 days before they could prepare fields for either transplanting or seeding with pre-germinated seed. Dry, direct seeding of rice contributes a valuable component of an adaptive strategy to address monsoonal variability that also may advance the time of wheat establishment and yield. Whilst the results illustrate the robustness, feasibility and significant potential of direct seeded rice, they also highlight the critical nature of effective weed control in successful implementation of direct seeding systems for rice.