In this paper I explore the question of what counts as successful implementation of a mathematics professional development (PD) project at scale. Two kinds of upscale are presented and differentiated from one another: upscale of settings and upscale of values, and the link between this differentiation and how success may be defined is discussed. As part of the discussion, three models of taking a PD project to scale are described, namely replication, adaptation and reproduction, and these are tied to the degree of professional autonomy of teachers within the PD. The examination of these ideas draws on data from two PD projects for secondary mathematics teachers in Israel, that are now years after their pilot phase and still being implemented. The first project is aimed at supporting mathematics teachers in teaching mathematically at-risk students and in helping these students to succeed in high-stakes exams. The second project is aimed at enhancing mathematics teachers' reflection on their teaching practices, using classroom video as a tool. Both projects were implemented at scale across Israel, and reviewing results from these implementations enables some general insights to be gained regarding the success of scaled-up PD projects. The crucial role of PD facilitators in successful implementations is highlighted.