BACKGROUND: To reduce maternal mortality Tanzania introduced Maternal Death Surveillance and Response (MDSR) system in 2015 as recommended by World Health Organization (WHO). All health facilities are to notify and review all maternal deaths inorder to recommend quality improvement actions to reduce deaths in future. The system relies on consistent and correct categorization of causes of maternal deaths and three phases of delays. To assess its adequacy we compared the routine MDSR categorization of causes of death and three phases of delays to those assigned by an independent expert panel with additional information from Verbal Autopsy (VA). METHODS: Our cross-sectional study included 109 reviewed maternal deaths from two regions in Tanzania for the year 2018. We abstracted the underlying medical causes of death and the three phases of delays from MDSR system records. We interviewed bereaved families using the standard WHO VA questionnaire. The obstetrician expert panel assigned underlying causes of death based on information from medical files and VA according to International Classification of Disease to Death in Pregnancy Childbirth and Puerperium (ICD-MM). They assigned causes to nine ICD-MM groups and identified the three phases of delays. We used Cohen's K statistic to compare causes of deaths and delays categorization. RESULTS: Comparison of underlying causes was done for 99 deaths. While 109 and 84 deaths for expert panel and MDSR respectively were analyzed for delays because of missing data in MDSR system. Expert panel and MDSR system assigned the same underlying causes in 64(64.6%) deaths (K statistic 0.60). Agreement increased in 80 (80.8%) when causes were assigned by ICD-MM groups (K statistic 0.76). The obstetrician expert panel identified phase one delays in 74 (67.9%), phase two in 24 (22.0%) and phase three delays in all 101 (100%) deaths that were assessed for this delay while MDSR system identified delays in 42 (50.0%), 10 (11.9%) and 78 (92.9%).The expert panel found human errors in management in 94 (93.1%) while MDSR system reported in 53 (67.9%) deaths. CONCLUSIONS: MDSR committees performed reasonably well in assigning underlying causes of death. The obstetrician expert panel found more delays than reported in MDSR system indicating difficulties within MDSR teams to critically review deaths.