Objectives: to analyze the spatial distribution of fetal deaths before and after implementation of surveillance for this event in the city of Recife, in the Northeast Region of Brazil. Methods: an ecological study whose spatial analysis unit was the 94 neighborhoods. The gross fetal mortality rates were calculated and the local empirical Bayesian estimator was adopted to smooth out random fluctuations of such rates. To analyze the spatial autocorrelation, the Global Moran’s Index was used, and spatial clusters were located by the Local Moran’s Index. Results: during the period before implementation of death surveillance, 1,356 fetal deaths were reported, a coefficient of 9.9 deaths per thousand births. During the second period, 1,325 fetal deaths occurred, a coefficient of 9.6 deaths per thousand births. The Global Moran’s Indexes (I) were I=0.6 and I=0.4 for the first and second periods, respectively, with statistical significance (p<0.05). For both periods analyzed, spatial clusters of high-risk neighborhoods were identified in the northern and eastern regions of the city. Conclusion: the spatial analysis indicated areas that persist as priorities for planning surveillance and health assistance actions to reduce fetal mortality.