When regulators are faced with practical challenges, policy instrument choice theories can help them find the best solution. However, not all such theories are equally helpful. This paper aims to offer regulators a better alternative to the current policy instrument choice theories. We will specifically address the shortcomings of “smart regulation theory” and present an alternative that keeps the best of that theory while remediating its weak points. Some authors (Böcher and Töller 2003; Baldwin and Black 2008) say that smart regulation theory does not address institutional issues, compliance type-specific response, performance-sensitivity and adaptability of regulatory regime. We have resolved these problems by merging the smart regulation theory with the policy arrangement approach and the policy learning concept. We call the resulting approach “regulatory arrangement approach” (RAA). The central idea of the RAA is to constrain the almost infinite “smart” regulatory options by: the national policy style; adverse effects of policy arrangements of adjoining policies; the structure of the policy arrangement of the investigated policy and competence dependencies of other institutions. The reduction can be so drastic that the potential governance capacity falls below the smart regulation threshold. In other words, no smart regulatory arrangement can be developed in that institutional context unless policy learning occurs. In addition, a “smart” regulatory arrangement is no guarantee that the policy will succeed. For this reason, the performance of the regulatory arrangement is measured and evaluated. Performance below a certain threshold indicates that the regulatory arrangement needs to be adapted, which then results in policy learning. We illustrate the usefulness of this new approach with a secondary analysis of the Flemish sustainable forest management policy.