Under the European chemicals legislation REACH (Registration, Evaluation, Authorisation and restriction of CHemicals), the use of chemicals posing an unacceptable risk for humans and the environment can be restricted. This requires that regulatory authorities of EU member states, or the European Chemicals Agency on request of the Commission, submit a restriction proposal in which they suggest one or multiple risk management options (RMOs). The options are recommended to be evaluated in a socio-economic analysis (SEA) using defined criteria. This paper explores the drivers of the selection of the preferred RMO in 32 restriction dossiers. Applying principal component analysis reveals that the selection of the preferred RMO, and the evaluation of possible trade-offs between alternative RMOs, is determined by criteria characterizing a measure's effectiveness and practicality, in particular its risk reduction capacity (R) and proportionality. A logistic regression using quantitative estimates provided in SEA suggests that the probability for an RMO to be selected is the higher the higher its R and the lower the costs of the restriction. Based on our analysis we conclude that the selection process of RMOs in REACH restriction dossiers could be strengthened by defining a limited but unambiguous set of criteria, conducting a score-based evaluation as a default, and by defining transparent decision rules.