BACKGROUND AND AIMS:In February 2014, pubs and bars in Kings Cross (KX) and the central business district (CBD) of Sydney, Australia were required to stop serving alcohol by 3 a.m. and operate a 'lockout' from 1.30 a.m. We studied changes in the incidence of assault during the following 5 years, including possible displacement. DESIGN:Controlled before-and-after ecological study. SETTING:Sydney, Australia. PARTICIPANTS/CASES:Assaults in specified areas in which police apprehended at least one person. MEASUREMENTS:We compared change in non-domestic assault in KX and the CBD with adjacent areas, other city areas, and outer suburbs, adjusting for the trend in the rest of the state during three periods: 6 p.m.-1.29 a.m. ('pre-lockout'), 1.30 a.m.-2.59 a.m. ('lockout') and 3 a.m.-6 a.m. ('after last-drinks'). We constructed interrupted time-series models with terms for secular trend and season, producing incidence rate ratios (IRR) for step and slope parameters. We performed sensitivity analyses on impacts of missing location data. FINDINGS:After the intervention, assaults fell 38% in KX (IRR for step change = 0.62, 95% CI = 0.49, 0.79) and 10% in the CBD (IRR = 0.90, 95% CI = 0.80, 0.99). Assaults continued declining in KX (IRR for slope = 0.990, 95% CI = 0.982, 0.998) and later increased in adjacent areas (IRR for slope = 1.006, 95% CI = 1.001, 1.011) and earlier in the evenings in both KX and the adjacent areas. The net reduction was 627 assaults over 60 months post-intervention, i.e. 10 fewer per month. Estimates were robust to extreme assumptions about missing data. CONCLUSIONS:The 2014 alcohol supply restrictions for pubs and bars in Kings Cross (KX) and the central business district (CBD) of Sydney, Australia were followed by a substantial reduction in the incidence of assault in KX and to a lesser extent in the CBD, possibly displacing some cases to adjacent areas and earlier in the evening.