INTRODUCTION: Pregnancy-related infection causes an estimated 11% of maternal deaths and increases the risk of stillbirth and neonatal mortality. However, definitions vary, measurement methods are inconsistent, and the incidence remains poorly described. This thesis aims to improve understanding of the measurement, incidence and risk factors of maternal peripartum infection. METHODS: I conducted a systematic literature review of global incidence of maternal peripartum infection; I explored infection definitions and data collection methods. I conducted a literature review of postnatal follow-up methods. Applying learning from the reviews, I designed a telephone-surveillance cohort study to measure incidence and risk factors of postnatal infection in Tanzania. RESULTS: No existing study met the full WHO criteria for maternal peripartum infection. In highquality studies, pooled infection incidence per 1000 women was 39 for chorioamnionitis, 16 for endometritis, 12 for wound infection and 0.5 for sepsis. Only 19% of studies met all quality criteria and 41% used a standard definition for infection. Less than half of studies followed women after hospital discharge. In the literature review of postnatal follow-up, telephone surveillance studies reached 63-91% of women. We recruited 879 women and interviewed 791 (90%) by telephone in Tanzania. Age, delivery mode and hospital did not affect the chance of reaching women, but 29% of interviews required over one call attempt. At day-28 postnatal, infection incidence per 1000 was 49 for maternal peripartum infection; 27 for endometritis, 28 for wound infection and with no cases of chorioamnionitis. The infection rate was higher in women with caesarean childbirth. CONCLUSION: Maternal peripartum infection remains an important complication of pregnancy and prevention strategies need increased attention. Improved measurement requires validated, standard definitions for constituent infections, applicable to low-resource settings, plus active postnatal follow-up. Telephone surveillance should be considered for follow-up; in Tanzania it achieved good coverage, and infection estimates were consistent with other studies.