The increasing numbers of workers in employment with little to no job security, so-called precarious employment, has led to a range of concerns over worker outcomes. A particular focus is the effect of instability on health in general, and particularly, mental health. We provide new evidence on this, focusing on an extreme form of precarious employment that has grown rapidly in the UK, zero-hours contracts (ZHCs). We demonstrate that workers employed on ZHCs are more likely to report a long-lasting health problem than workers employed on other types of contract. In particular, reported levels of mental ill health are higher (almost double) among ZHC workers than for other workers. These associations remain, and a positive association between ZHC employment and physical ill health emerges, after controlling for a rich set of observable characteristics. Estimated associations vary little between different demographic groups, although they are concentrated in parts of the economy where underlying job instability is likely to be higher. Finally, we exploit sectoral variation in the historical prevalence of ZHC-like employment, in an instrumental variables framework, to demonstrate large and potentially causal effects of ZHC employment on reporting a long-lasting health problem and on mental ill health, but no effect on physical health. It is unlikely that these effects are currently factored into short-term employment conditions or that they attract compensating wage differentials.