OBJECTIVE: Modelling the relationship between weather, climate and infectious diseases can help identify high-risk periods and provide understanding of the determinants of longer-term trends. We provide a detailed examination of the non-linear and delayed association between temperature and salmonellosis in three New Zealand cities (Auckland, Wellington and Christchurch). METHODS: Salmonella notifications were geocoded to the city of residence for the reported case. City-specific associations between weekly maximum temperature and the onset date for reported salmonella infections (1997-2007) were modelled using non-linear distributed lag models, while controlling for season and long-term trends. RESULTS: Relatively high temperatures were positively associated with infection risk in Auckland (n=3,073) and Christchurch (n=880), although the former showed evidence of a more immediate relationship with exposure to high temperatures. There was no significant association between temperature and salmonellosis risk in Wellington. CONCLUSIONS: Projected increases in temperature with climate change may have localised health impacts, suggesting that preventative measures will need to be region-specific. This evidence contributes to the increasing concern over the public health impacts of climate change.