In November 2016, the UK government announced they would be lowering the benefit cap, the total amount a family with no-one in full-time employment can receive from the government in social security. This policy change reduced financial support for large (often lone parent) families and those with high housing costs, and broke the link between needs and entitlements in the British social security system. This policy was intended to incentivise people to return to work but it may have also harmed mental health, especially because those affected by the reform may struggle to find appropriate work or move to cheaper housing. We treat this reform as a natural policy experiment, comparing those at-risk of being capped and those who were not, and then examining the risk of experiencing poor mental health both before and after the cap was lowered. The main outcome is a binary measure of self-reported mental health problems. Drawing on data from 1.4 million individuals collected between January 2015 and December 2018, we find that the prevalence of depression or anxiety among those at-risk of being capped increased by 2.6 percentage points (95% confidence interval: 1.33 to 3.88) compared with those at a low risk of being capped. This association is consistent to a variety of sensitivity tests. We conclude that lowering the total amount of financial assistance families can receive in social security may increase the risk of mental ill health and could have the unintended consequence of pushing out-of-work people even further away from the labour market.