Various scores attempt to predict the development of RA. In particular, EULAR proposes a simple rule to identify new onset arthralgias suspicious of progression to RA. However, serological tests are necessary to ensure acceptable specificity. In patients with clinical arthritis, reliable predictive criteria for progression to RA have also been identified. Overall, the validity of the available scores is still debated. Such scores limitedly take into account interactions between risk factors in specific subpopulations. New technologies could help to overcome these limitations, but we need databases containing sufficient numbers of RA and pre-RA patients, including pre-diagnostic follow-up. Today, existing predictive rules do not seem to compete significantly with expert opinions.