Objectives: to identify factors associated with homicides in women who had prior notification of violence in the state of Pernambuco, 2011 to 2016. Methods: a transversal, quantitative and observational study, guided by the STROBE tool, with data of homicides of women obtained from the Mortality Information System; and of violence against women, from the Grievance Notification Information System. The logistic regression model was used to verify the factors associated with the homicides. Results: there were 32,308 cases of violence against women and 1,162 homicides. The chances of homicide were higher for women: victims of physical violence (2.39 times more), aggression by object of court (2.32 times more), aggression by firearm (6 times more), and when there was a recurrence of violence (3.82 times more). Conclusions: association of physical violence, agression by object of court, aggression by firearm, recurrence of violence with homicides of women in the state of Pernambuco was found.