Abstract We present an experimental method for modelling lithospheric deformation. The method is based on a simplified profile of lithosphere rheology where only four layers are considered: brittle crust, ductile crust, brittle mantle and ductile mantle. The strengths of the four layers, calculated from laboratory determined flow laws, depend strongly on geotherms and slightly on strain rates and deformation regimes. The rheological profiles of the lithosphere have been recreated in experimental models built with analog materials (sand, silicone putty and golden syrup) which flow under their own weight. During lithospheric shortening, the style of deformation depends on the relative strengths of the four layers, especially those in the mantle. Of special importance is the existence, or non-existence, of a brittle mantle. The initial stages of compression often produce flexures or buckles without local isostatic compensation. Buckling exerts a strong control on the localisation of thrust faults in later stages: the thrusts tend to form at the inflexion points of buckles. The style of crustal thickening depends on the relative thicknesses and resistances of the lithospheric layers. In 2-layer lithospheric models (without resistant mantle), shortening was concentrated upon a small number of thrusts, widely spaced. In 3-layer models, shortening was concentrated into a large number of reverse faults more closely spaced and more steeply dipping than before. In 4-layer models, a single thrust formed in the brittle mantle. Motion on this thrust dragged the crustal layers down, forming an asymmetric crustal root with steep imbricate slices.