Terminal drought during the reproductive stage is a major constraint to yield of chickpea in many regions of the world. Termination of watering (WS) during podding in a small-seeded desi chickpea (Cicer arietinum L.) cultivar, Rupali, and a large-seeded kabuli chickpea cultivar, Almaz, induced a decrease in predawn leaf water potential (LWP), in the rate of photosynthesis, and in stomatal conductance. Compared to well-watered (WW) controls, the WS treatment reduced flower production by about two-thirds. In the WW treatment, about 15% of the flowers aborted and 42% (Rupali) and 67% (Almaz) of the pods aborted, whereas in the WS treatment 37% and 56% of the flowers aborted and 54% and 73% of the pods aborted, resulting in seed yields of 33% and 15% of the yields in WW plants in Rupali and Almaz, respectively. In vitro pollen viability and germination in Rupali decreased by 50% and 89% in the WS treatment, and pollen germination decreased by 80% in vivo when pollen from a WS plant was placed on a stigma of a WW plant. While about 37% of the germinated pollen tubes from WW plants and 22% from the WS plants reached the ovary in the WW plants, less than 3% of pollen grains reached the ovary when pollen from either WS or WW plants was placed on a stigma of a WS plant. It is concluded that, in addition to pod abortion, flower abortion is an important factor limiting yield in chickpea exposed to terminal drought and that water deficit impaired the function of the pistil/style more than the pollen.