Background and purpose To establish post-stroke case fatality rates within a community based incident stroke population in rural Tanzania.

Methods Incident stroke cases were identified by the Tanzanian Stroke Incidence Project and followed-up over the next 3–6 years. In order to provide a more complete picture, verbal autopsy (VA) was also used to identify all stroke deaths occurring within the same community and time period, and a date of stroke was identified by interview with a relative or friend.

Results Over 3 years, the Tanzanian Stroke Incidence Project identified 130 cases of incident stroke, of which 31 (23.8%, 95% CI 16.5 to 31.2) died within 28 days and 78 (60.0%, 95% CI 51.6 to 68.4) within 3 years of incident stroke. Over the same time period, an additional 223 deaths from stroke were identified by VA; 64 (28.7%, 95% CI 20.9 to 36.5) had died within 28 days of stroke and 188 (84.3%, 95% CI 78.1 to 90.6) within 3 years.

Conclusions This is the first published study of post-stroke mortality in sub-Saharan Africa from an incident stroke population. The 28 day case fatality rate is at the lower end of rates reported for other low and middle income countries, even when including those identified by VA, although CIs were wide. Three year case fatality rates are notably higher than seen in most developed world studies. Improving post-stroke care may help to reduce stroke case fatality in sub-Saharan Africa.