OBJECTIVE: To assess stillbirth mortality by Robson ten-group classification and the usefulness of this approach for understanding trends. DESIGN: Cross-sectional study. SETTING: Prospectively collected perinatal e-registry data from 16 hospitals in Benin, Malawi, Tanzania and Uganda. POPULATION: All women aged 13-49 years who gave birth to a live or stillborn baby weighting >1000 g between July 2021 and December 2022. METHODS: We compared stillbirth risk by Robson ten-group classification, and across countries, and calculated proportional contributions to mortality. MAIN OUTCOME MEASURES: Stillbirth mortality, defined as antepartum and intrapartum stillbirths. RESULTS: We included 80 663 babies born to 78 085 women; 3107 were stillborn. Stillbirth mortality by country were: 7.3% (Benin), 1.9% (Malawi), 1.6% (Tanzania) and 4.9% (Uganda). The largest contributor to stillbirths was Robson group 10 (preterm birth, 28.2%) followed by Robson group 3 (multipara with cephalic term singleton in spontaneous labour, 25.0%). The risk of dying was highest in births complicated by malpresentations, such as nullipara breech (11.0%), multipara breech (16.7%) and transverse/oblique lie (17.9%). CONCLUSIONS: Our findings indicate that group 10 (preterm birth) and group 3 (multipara with cephalic term singleton in spontaneous labour) each contribute to a quarter of stillbirth mortality. High mortality risk was observed in births complicated by malpresentation, such as transverse lie or breech. The high mortality share of group 3 is unexpected, demanding case-by-case investigation. The high mortality rate observed for Robson groups 6-10 hints for a need to intensify actions to improve labour management, and the categorisation may support the regular review of labour progress.