Water availability plays a major role in the regulation of seed development. Experimental evidence on the water relations of seed development is reviewed, with emphasis on the measured water potential values of well-watered and water-stressed seed, fruit, and leaf tissues. The hypothesis of hydraulic isolation of developing seeds from the water relations of the maternal plant, proposed on the basis of the maintenance of measured water potential (psi(w)) differences between these tissues, is examined and found to be problematic. Apparent psi(w) gradients have also been recorded for seeds developing within fleshy fruits or in vitro on defined medium, where hydraulic barriers would not be present. Current models of phloem transport to seeds are based on a turgor homeostatic mechanism in the unloading tissues that maintains high apoplastic solute concentrations to lower phloem turgor and promote import. The combination of turgor (or water potential) homeostasis and active regulation of solute compartmentation in seed tissues may contribute to the apparent constancy of seed psi(w) when measured by thermocouple psychrometry, which requires long equilibration times after excision. The delivery of excess water to developing seeds via the phloem and its recirculation to the mother plant also present problems for the hydraulic resistance hypothesis and for the retention of apoplastic solutes in the unloading zone. It is hypothesized that modfied cell walls acting as semipermeable apoplastic membranes may retain solutes within the unloading tissues, while allowing the return flow of water to the parental xylem. The water relations of developing seeds are determined primarily by phloem transport, whether under well-watered or water-stressed conditions, so future studies should attempt to resolve the discrepancies between measured psi(w) and the requirements for maintenance of phloem import.