There is a strong positive correlation between bathymetric and well-navigated heat flow data from the Galapagos Mounds hydrothermal area and the east flank of the Juan de Fuca Ridge. A similar correlation is apparent at Deep Sea Drilling Project/Ocean Drilling Program (DSDP/ODP) site 504, where it provides a constraint on the intensity and geometry of ridge flank hydrothermal circulation. We test the implications of this correlation with a hydrogeological model of the shallow crust that contains significant permeability mainly within a few thin layers (tens of meters thick) in the upper few hundred meters of basement, consistent with observations in DSDP/ODP holes into upper oceanic crust. These simulations are compared directly to others in which the same bulk permeability is represented more homogeneously within the upper crust. A model with thin permeable zones and subtle relief beneath a flat seafloor can explain the variations in seafloor heat flow observed on one part of the Juan de Fuca Ridge east flank, as can a model with a flat seafloor and basement top but relief within the most permeable zone. We explore the importance of layered heterogeneities versus explicit permeability anisotropy in these simulations through a series of parametric tests. Some form of permeability anisotropy appears to be required in order to achieve efficient lateral heat transport within the upper volcanic crust so as to produce the common seafloor heat flow-basement relief correlation. This permeability anisotropy may result directly from the primary architecture of the volcanic crust. Numerical results suggest that within an anisotropic system such as the upper oceanic crust, the length scale of heat flow variations (and therefore the length scale of underlying hydrothermal convection cells) is not an indication of the depth extent of fluid flow. Rather, the length scale of heat flow anomalies may reflect the length scale of relief along aquifer boundaries, while the depth of vigorous flow is constrained by the depth extent of significant permeability, and may be unresolvable with surface heat flow measurements. ¿ American Geophysical Union 1995