Accurate prediction of surface subsidence due to the extraction of underground coal seams is a significant challenge in geotechnical engineering. This task is further compounded by the growing trend for coal to be extracted from seams either above or below previously extracted coal seams, a practice known as multi-seam mining. In order to accurately predict the subsidence above single and multi-seam longwall panels using numerical methods, constitutive laws need to appropriately represent the mechanical behaviour of coal measure strata. The choice of the most appropriate model is not always straightforward. This paper compares predictions of surface subsidence obtained using the finite element method, considering a range of well-known constitutive models. The results show that more sophisticated and numerically taxing constitutive laws do not necessarily lead to more accurate predictions of subsidence when compared to field measurements. The advantages and limitations of using each particular constitutive law are discussed. A comparison of the numerical predictions and field measurements of surface subsidence is also provided.