The 1999 severe crop season drought in the northeastern US was followed by hurricane-driven torrential rains in September, offering a unique opportunity to observe how managed and natural systems respond to climate-related stress. The Rodale Institute Farming Systems Trial has been operating since 1981 and consists of three replicated cropping systems, one organic manure based (MNR), one organic legume based (LEG) and a conventional system (CNV). The MNR system consists of a 5-year maize–soybean–wheat–clover/hay rotation, the LEG of a 3-year maize–soybean–wheat–green manure, and the CNV of a 5-year maize-soybean rotation. Subsoil lysimeters allowed quantification of percolated water in each system. Average maize and soybean yields were similar in all three systems over the post-transition years (1985–1998). Five drought years occurred between 1984 and 1998 and in four of them the organic maize outyielded the CNV by significant margins. In 1999 all crop systems suffered severe yield depressions; however, there were substantial yield differences between systems. Organic maize yielded 38% and 137% relative to CNV in the LEG and MNR treatments, respectively, and 196% and 152% relative to CNV in the soybean plots. The primary mechanism of the higher yield of the MNR and LEG is proposed to be the higher water-holding capacity of the soils in those treatments, while the lower yield of the LEG maize was due to weed competition in that particular year and treatment. Soils in the organic plots captured more water and retained more of it in the crop root zone than in the CNV treatment. Water capture in the organic plots was approximately 100% higher than in CNV plots during September's torrential rains.