Two competing models have been proposed for Early Jurassic magmatism on the eastern North American margin. The "broad terrane" hypothesis argues that tholeiitic lavas were extruded over a large area and later eroded. Alternatively, the lavas were extruded only in the basins in which they now outcrop. We compare the stratigraphy and geochemistry of the tholeiitic lavas and dykes from Atlantic Canada with those of the type section of the Newark basin and use this correlation to test these two models. The earliest high-Ti quartz tholeiites in the type section are represented by lavas in the Fundy basin (North Mountain Basalt), Scotian basin, and eastern Grand Banks and by the Shelburne and Ministers Island dykes. Spatial and temporal geochemical variations in the North Mountain Basalt are mirrored by the Shelburne dyke, strengthening the case that the two were geochemically related. Basalts in Grand Manan Island, on the footwall of the Grand Manan fault that bounds the Fundy basin, are geochemically similar to the lowest North Mountain Basalt flow. These observations suggest that the earliest basalt flows were originally more extensive and have become restricted by later uplift and erosion. However, younger magmas in the Newark basin are represented only by the Caraquet, Anticosti, and Avalon dykes in Atlantic Canada, and corresponding lavas were never deposited in the Fundy basin or eastern Grand Banks. Thus, Jurassic tholeiitic lava distribution lies in between the predictions of the "broad terrane" and the "restricted basin" models.