Background New Zealand is a temperate country with substantial excess winter mortality. We investigated whether this excess winter mortality varies with social factors.

Methods Records from New Zealand censuses in 1981, 1986, 1991, 1996 and 2001 were each anonymously and probabilistically linked to 3 years of subsequent mortality data creating five cohort studies of the New Zealand adult population (age 30–74 years at census) each with 3 years' follow-up. Logistic regression analysis was used to model the risk of dying in winter compared to summer with winter deaths classified ‘1’ and summer deaths ‘0’. There were 75 138 eligible mortality records with complete data on social variables recorded for 58 683 (78%).

Results Adjusting for age, sex, census year, ethnicity and tenure, those in the lowest tertile of income were at increased risk of winter death compared to those in the highest tertile: OR 1.13 (95% CI 1.08 to 1.19). Compared to home owners, people living in rented accommodation were at greater risk of winter death: OR 1.05 (95% CI 1.01 to 1.10). Urban dwellers were also at significantly increased risk. The strongest associations were seen for infectious diseases.

Conclusions There was an increased risk of dying in winter for most New Zealanders, but more so among low-income people, those living in rented accommodation and those living in cities. Exact causal mechanisms are not known but possibly include correlated poorer health status, low indoor temperatures and household crowding.