Some materials can at low temperatures be described by a spin Hamiltonian with dipolar interactions. Monte Carlo simulation is a method well suited to extract the temperature dependence of magnetic quantities from such models. The angular dependence and long range of the interactions give these systems interesting properties, which have been observed in experiments on LiHoF4 and related materials. Obtaining accurate enough numerical data to enable a comparison poses a serious computational challenge, in particular close to the phase transition. In this talk I will also mention something about how partially substituting non-magnetic Y for magnetic Ho affects the phase transition.