In recent years, both in social science and policy circles, there has been a growing consensus on the multidimensional nature of poverty. However, the operationalisation of the concept has not followed this development, as most studies are still primarily based on income. In this article, we propose to measure the concept of poverty using both monetary and non-monetary indicators. To this end, a latent class measurement model is used, allowing us to take account of the multidimensionality of the data and the discrete nature of most available poverty indicators. The proposed measurement instrument allows for poverty to manifest itself in different ways or forms for different subgroups in the population, Furthermore, the occurrence of multiple deprivation, rather than a 'negative' score on only one indicator, is taken into account. In addition, special attention is paid to the feasibility of constructing a multidimensional poverty measure which can be used to study poverty dynamics with longitudinal panel data. We present figures on the size and the social distribution of the 'poor' population in Belgium and Britain. These figures indicate that the results of our multidimensional measurement procedure are both plausible and substantively interpretable.