Whether it is better to adopt voice or exit depends on the nature of the transaction for both parties. The strongest voice-sustaining equilibrium is where both parties see voice as preferable to exit. This is likely where both parties have substantial sunk costs. In other circumstances, there is no voice-sustaining equilibrium. We apply insights from transaction cost economics to voice and exit in the employment relationship. We consider which types of employment relationship are likely to sustain voice and which are not using a model of the employment contract that originates with Oliver Williamson. Based on this, we use a simple cost benefit model which deals with two related questions. First, where is voice likely to emerge and, second, what type of voice is it likely to be? We then apply the model empirically to workplace data for Britain to shed light on the size and composition of the "no voice" sector; the emergence of a sizeable "non-union" voice sector; the persistence of unionization among existing establishments; and the economic outcomes associated with voice choice by firms.