The hypothesis that ABA produced by roots in drying soil is responsible for stomatal closure was tested with grafted plants constructed from the ABA-deficient tomato mutants, sitiens and flacca and their near-isogenic wild-type parent. Three types of experiments were conducted. In the first type, reciprocal grafts were made between the wild type and sitiens or flacca. Stomatal conductance accorded with the genotype of the shoot, not the root. Stomates closed in all of the grafted plants in response to soil drying, regardless of the root genotype, i.e. regardless of the ability of the roots to produce ABA. In the second type of experiment, wild-type shoots were grafted onto a split-root system consisting of one wild-type root grafted to one mutant (flacca or sitiens) root. Water was withheld from one root system, while the other was watered well so that the shoots did not experience any decline in water potential or loss of turgor. Stomates closed to a similar extent when water was withheld from the mutant roots or the wild-type roots. In the third type of experiment, grafted plants with wild-type shoots and either wild-type or sitiens roots were established in pots that could be placed inside a pressure chamber, and the pressure increased as the soil dried so that the shoots remained fully turgid throughout. Stomates closed as the soil dried, regardless of whether the roots were wild type or sitiens. These experiments demonstrate that stomatal closure in response to soil drying can occur in the absence of leaf water deficit, and does not require ABA production by roots. A chemical signal from roots leading to a change in apoplastic ABA levels in leaves may be responsible for the stomatal closure.