Sap flow gauges were used to estimate whole-plant water flux for five stem-diameter classes of red maple (Acer rubrum L.) growing in the understory of an upland oak forest and exposed to one of three large-scale (0.64 ha) manipulations of soil water content. This Throughfall Displacement Experiment (TDE) used subcanopy troughs to intercept roughly 30% of the throughfall on a "dry" plot and a series of pipes to move this collected precipitation across an "ambient" plot and onto a "wet" plot. Saplings with a stem diameter larger than 10 cm lost water at rates 50-fold greater than saplings with a stem diameter of 1 to 2 cm (326 versus 6.4 mol H(2)O tree(-1) day(-1)). These size-class differences were driven largely by differences in leaf area and cross-sectional sapwood area, because rates of water flux expressed per unit leaf area (6.90 mol H(2)O m(-2) day(-1)) or sapwood area (288 mol H(2)O dm(-2) day(-1)) were similar among saplings of the five size classes. Daily and hourly rates of transpiration expressed per unit leaf area varied throughout much of the season, as did soil matrix potentials, and treatment differences due to the TDE were observed during two of the seven sampling periods. On July 6, midday rates of transpiration averaged 1.88 mol H(2)O m(-2) h(-1) for saplings in the "wet" plot, 1.22 mol H(2)O m(-2) h(-1) for saplings in the "ambient" plot, and 0.76 mol H(2)O m(-2) h(-1) for saplings in the "dry" plot. During the early afternoon of August 28, transpiration rates were sevenfold lower for saplings in the "dry" plot compared to saplings in the "wet" plot and 2.5-fold lower compared to saplings in the "ambient" plot. Treatment differences in crown conductance followed a pattern similar to that of transpiration, with values that averaged 60% lower for saplings in the "dry" plot compared to saplings in the "wet" plot and 35% lower compared to saplings in the "ambient" plot. Stomatal and boundary layer conductances were roughly equal in magnitude. Estimates of the decoupling coefficient (Omega) ranged between 0.64 and 0.72 for saplings in the three TDE treatment plots. We conclude that red maple saplings growing in the understory of an upland oak forest are responsive to their edaphic and climatic surroundings, and because of either their small stature or their shallow root distribution, or both, are likely to be impacted by precipitation changes similar to those predicted by global climate models.