We have explored the idea that observations of large igneous crustal thickness at volcanic rifted continental margins may be explained by the interaction of rift-driven flow in the lithosphere with an underlying, sublithospheric hot plume sheet. This concept is assessed by numerical modeling of the viscous flow caused by divergent plate motions, using a viscous, temperature- and pressure-dependent, nonlinear fluid. The plume sheet consists of an initially hot layer just below the lithosphere which responds to the overlying lithospheric motions. For reasonable values of input parameters we can successfully predict the observed thickness of igneous crust at the margin, its formation time, inferred average mantle melt fractions, and, importantly, the eventual transition to normal oceanic crustal thicknesses. Observations are selected from the North Atlantic region, with an emphasis on the recent results from the east Greenland volcanic margin. The best fitting models suggest that relatively little small-scale convection is needed after the initial strong pulse of mantle upflow due to the presence of moderately hot (~1450 ¿C) buoyant plume material under the rift. The results favor a thin (~50 km) plume sheet layer. This hot mantle reservoir is soon depleted and replaced by normal temperature mantle, unless the rift lies directly over the deep source of plume material. ¿ 2000 American Geophysical Union