It is often claimed that the many worlds theory is to be preferred over other realist interpretations of quantum mechanics for its ability to avoid the kind of action at a distance that plagues both hidden variables and collapse models. The aim of this paper is address the question of whether branching should be viewed as a causal process that spreads out from a localized region as some authors (Wallace (2012), Blackshaw, Huggett, and Ladyman (manuscript)) have such suggested, or whether (...) it occurs globally across the total quantum state at an instant as others insist (Sebens and Carroll (2018), McQueen and Vaidman (2019)). An appeal of the local branching model is it that it ensures that branching never involves superluminal influences. After clarifying these two models, the paper will consider the argument in favor of local branching, and then four concerns one might have about that model, that speak in favor of regarding branching as global and instantaneous. Ultimately, although the first three concerns may be addressed by advocates of local branching, the fourth is shown to be more convincing, and speaks in favor of regarding branching as a global, instantaneous change, even if it is a change that is triggered by a local, causal process (decoherence). The last part of paper clarifies the proposed model of branching as global, and explains how it is not in tension with relativity or the many worlds interpretations’ claim to avoid spooky action at a distance. (shrink)