BACKGROUND: Maternal and neonatal mortality remains a serious challenge in Tanzania. Progress is tracked through maternal mortality ratios (MMR) and neonatal mortality rates (NMR), yet robust national data on these outcomes is difficult and expensive to ascertain, and mask wide variation. SEARCH STRATEGY: We searched EMBASE, MEDLINE, Popline, and EBSCO online databases, basing search terms on ("maternal" OR "neonatal") AND ("mortality" OR "cause of death") AND "Tanzania." SELECTION CRITERIA: Nationally representative or population representative from the subnational context were eligible, providing NMR, MMR, or numbers of maternal deaths or early neonatal deaths or neonatal deaths and live births. DATA COLLECTION AND ANALYSIS: Data were extracted on study context, time period, number of deaths and live births, definition of maternal and neonatal death, study design, and completeness and representativeness of data. NMR and MMR were extracted or calculated and study quality was assessed. Nationally representative data were compared with modelled national data from international agencies. MAIN RESULTS: 2107 records were screened yielding 21 maternal mortality and 15 neonatal mortality datasets. There were high mortality levels with wide subnational MMR and NMR variation. National survey data differed from the modelled estimates, with wide uncertainty ranges. CONCLUSION: Subnational data quality was generally poor with no observable trends and geographical clustering across several regions. Combined MMR and NMR reporting is uncommon. Modelled national estimates lack precision and are complex to interpret. Results suggest that aggregate national data are inadequate for policy generation and progress monitoring. We recommend strengthening of vital registration and Health Management Information Systems with complementary use of process indicators, for improved monitoring of, and accountability for maternal and newborn health.