Irrigation represents one of the main strategies to enhance and stabilize agricultural productivity, by mitigating the effects of rainfall vagaries. In the face of the projected growth in population and in biofuel demands, as well as shifts in climate and dietary habits, a more sustainable management of water resources in agroecosystems is needed. The field of ecohydrology, traditionally focusing on natural ecosystems, has the potential to offer the necessary quantitative tools to assess and compare agricultural enterprises across climates, soil types, crops, and irrigation strategies, accounting for the unpredictability of the hydro-climatic forcing. Here, agricultural sustainability and productivity are assessed with reference to water productivity (defined as the ratio between yield and total supplied water), yields, water requirements, and their variability-a crucial element for food security and resource allocation planning. These synthetic indicators are quantified by means of a probabilistic description of the soil water balance and crop development. The model results allow the interpretation of patterns of water productivity observed in Zea mays (maize) and Triticum aestivum (wheat), grown under a variety of soils, climates, and irrigation strategies. Employing the same modeling framework, the impact of rainfall pattern and irrigation strategy on yield and water requirements is further explored. The obtained standard deviations of yield and water requirements suggest the existence of a nonlinear tradeoff between yield stabilization and variability of water requirements, which in turn is strongly impacted by irrigation strategy. Moreover, intermediate rainfall amounts are associated to the highest variability in yields and irrigation requirements, although allowing the maximum water productivity. The existence of these tradeoffs between productivity, reliability, and sustainability poses a problem for water management, in particular in mesic climates.