Abstract Alcohol-related violence is a pressing public health concern. In 2005, the government of England and Wales took a controversial approach to preventing violence by removing restrictions on opening hours for alcohol outlets, thus increasing the availability of alcohol. The policy aimed to remove fixed closing times, which it claimed was contributing to urban violence occurring at peak closing times. It proposed to reduce violence and disorder by installing systems of ‘staggered closing times’. This policy was criticised for overlooking established public health principles prioritising the control of alcohol availability in the prevention of alcohol-related harm. In this study, we treated the removal of trading hour restrictions as a natural experiment to test competing theoretical principles about the relationship between alcohol availability and violence. Our study took place in the City of Manchester over a four-year period 2004–2008. Detailed trading records for over 600 alcohol outlets were obtained, as were police records for all violent incidents. We found considerable variation in the implementation of extended trading hours across the city, which affected area-level exposure of changes in alcohol availability and staggered closing times. To isolate the effect of these changes on violence, we performed a dose–response analysis to examine whether improved staggering of closing hours (or increased alcohol availability) was associated with decreases in violence. We found no evidence to support the government-proposed hypothesis that staggered closing reduces violence. We also found no support for the alternative hypothesis; that increase alcohol availability would result in increased violence. This study provides an example of how better evidence can be generated from natural experiments by placing added emphasis on theory, causal mechanisms and implementation science.