The hydraulic limitation hypothesis (Ryan and Yoder 1997) proposes that leaf-specific hydraulic conductance (kl) and stomatal conductance (gs) decline as trees grow taller, resulting in decreased carbon assimilation. We tested the hydraulic limitation hypothesis by comparison of canopy-dominant Douglas-fir (Pseudotsuga menziesii var. menziesii (Mirb.) Franco) trees in stands that were approximately 15 m (20 years old), 32 m (40 years old) and 60 m (> 450 years old) tall in Wind River, Washington, USA. Carbon isotope discrimination (Delta) declined with tree height (18.6, 17.6 and 15.9 per thousand for stands 15, 32 and 60 m tall, respectively) indicating that gs may have declined proportionally with tree height in the spring months, when carbon used in the construction of new foliage is assimilated. Hydraulic conductance decreased by 44% as tree height increased from 15 to > 32 m, and showed a further decline of 6% with increasing height. The general nonlinear pattern of kl versus height was predicted by a model based on Darcy's Law. Stemwood growth efficiency also declined nonlinearly with height (60, 35 and 28 g C m-2 leaf area for the 15-, 32- and 60-m stands, respectively). Unlike kl and growth efficiency, gs and photosynthesis (A) during summer drought did not decrease with height. The lack of decline in cuvette-based A indicates that reduced A, at least during summer months, is not responsible for the decline in growth efficiency. The difference between the trend in gs and A and that in kl and D may indicate temporal changes (spring versus summer) in the response of gas exchange to height-related changes in kl or it may be a result of measurement inadequacies. The formal hydraulic limitation hypothesis was not supported by our mid-summer gs and A data. Future tests of the hydraulic limitation hypothesis in this forest should be conducted in the spring months, when carbon uptake is greatest. We used a model based on Darcy's Law to quantify the extent to which compensating mechanisms buffer hydraulic limitations to gas exchange. Sensitivity analyses indicated that without the observed increases in the soil-to-leaf water potential differential (DeltaPsi) and decreases in the leaf area/sapwood area ratio, kl would have been reduced by more than 70% in the 60-m trees compared with the 15-m trees, instead of the observed decrease of 44%. However, compensation may have a cost; for example, the greater DeltaPsi of the largest trees was associated with smaller tracheid diameters and increased sapwood cavitation, which may have a negative feedback on kl and gs.