The large thickness of Upper Carboniferous strata found in the Netherlands suggests that the area was subject to long-term subsidence. However, the mechanisms responsible for subsidence are not quantified and are poorly known. In the area north of the London Brabant Massif, onshore United Kingdom, subsidence during the Namurian-Westphalian B has been explained by Dinantian rifting, followed by thermal subsidence. In contrast, south and east of the Netherlands, along the southern margin of the Northwest European Carboniferous Basin, flexural subsidence caused the development of a foreland basin. It has been proposed that foreland flexure due to Variscan orogenic loading was also responsible for Late Carboniferous subsidence in the Netherlands. In the first part of this paper, we present a series of modelling results in which the geometry and location of the Variscan foreland basin was calculated on the basis of kinematic reconstructions of the Variscan thrust system. Although several uncertainties exist, it is concluded that most subsidence calculated from well data in the Netherlands cannot be explained by flexural subsidence alone. Therefore, we investigated whether a Dinantian rifting event could adequately explain the observed subsidence by inverse modelling. The results show that if only a Dinantian rifting event is assumed, such as is found in the United Kingdom, a very high palaeowater depth at the end of the Dinantian is required to accommodate the Namurian-Westphalian B sedimentary sequence. To better explain the observed subsidence curves, we propose (1) an additional stretching event during the Namurian and (2) a model incorporating an extra dynamic component, which might well explain the very high wavelength of the observed subsidence compared with the wavelength of the predicted flexural foreland basin. © 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd.