Objective To assess the acute effects of daily mean temperature,cold spells,and heat waves on stroke mortality in 12 counties across Hubei province,China.Methods Data related to daily mortality from stroke and meteorology in 12 counties across Hubei province during 2009-2012,were gathered.Distributed lag nonlinear model (DLNM) was first used,to estimate the county-specific associations between daily mean temperature,cold spells,heat waves and stroke mortality.Multivariate Meta-analysis was then applied to pool the community-specific relationships between temperature and stroke mortality (exposure-response relationship) as well as both cold-and-heatassociated risks on mortality at different lag days (lag-response relationship).Results During 2009-2012,a total population of 6.7 million was included in this study with 42 739 persons died of stroke.An average of 2.7 (from 0.5 to 6.0) stroke deaths occurred daily in each county,with annual average mean temperature as 16.6 ℃ (from 14.7 ℃ to 17.4 ℃)during the study period.An inverse J-shaped association between temperature and stroke mortality was observed at the provincial level.Pooled mortality effect of cold spells showed a 2-3-day delay and lasted about 10 days,while effect of heat waves appeared acute but attenuated within a few days.The mortality risks on cold-spell days ranged from 0.968 to 1.523 in 12 counties at lag 3-14,with pooled effect as 1.180 (95％ CI:1.043-1.336).The pooled mortality risk (ranged from 0.675 to 2.066) on heat-wave days at lag 0-2 was 1.114 (95％CI:1.012-1.227).Conclusions An inverse J-shaped association between temperature and stroke mortality was observed in Hubei province,China.Both cold spells and heat waves were associated with increased stroke mortality,while different lag patterns were observed in the mortality effects of heat waves and cold spells.