The concept of behavioural additionality, which was originally associated with the question of "what difference does policy make in the behaviour of the firms it supports?", has increasingly been used by scholars, evaluators and policy-makers not only in the field of evaluation of innovation policies but also within the context of fundamental discussions of policy rationales. A survey of the evaluation literature concludes that the concept lacks a clear definition and theoretical background as it does not have a unit of analysis. An empirical analysis of the current practice of innovation policy evaluation endorses this finding and outlines the uniqueness of behavioural additionality in terms of its evaluation. A survey of policy discussions around the concept reveals that behavioural additionality has been argued as the hallmark of an evolutionary view in respect of innovation policy. However, the thesis suggests that the concept also lacks an appropriate framework of analysis to accomplish this.The thesis argues that behavioural additionality should be redefined by using the concept of organisational routines as the unit of analysis and the evolutionary approach as the framework of analysis. To this end, a theoretical and generic approach that conceives behavioural additionality as the government-influenced evolution of organisational routines at the micro, meso and macro levels is devised. The thesis unfolds how behavioural additionality is created by reinforcing ostensive, performative and artefact aspects of routines within firms. The possibility, rationale and evaluation of behavioural additionality are also discussed. Empirically, the thesis applies a plausibility probe that employs two case studies of Turkish TIDEB and British Collaborative R&D programmes to illustrate the micro level of the approach developed.It is concluded that the proposed approach provides a better theoretical understanding for behavioural additionality, which would increase its impact on policy-making. This new approach also represents a concrete attempt to utilise the framework and unit of analysis of the evolutionary approach in the field of evaluation for the first time.