Abstract. We assessed changes in the effect size of risk factors for infant mortality comparing a birth cohort from 2005 to 2010 with a birth cohort from 1993 to 1999 in the Nouna Health and Demographic Surveillance System (HDSS) in Burkina Faso. Single-and three-level Cox proportional hazards regression models were used for analysis. Independent variables among others included year of birth, ethnicity, religion, age of the mother, birth order, death of the mother, being a twin, and distance to the closest health facility. We observed an infant mortality rate of about 51/1,000 person-years. The strongest risk factors were death of the mother and being a twin, which were also the strongest risk factors from the previous analysis period. Compared with the period 1993-1999, the effect of most risk factors decreased, notably ethnicity, religious affiliation, distance to the closest health facility, birth order, and season of birth. The strongest reduction in mortality occurred in the groups with the previously highest infant mortality rates in 1993-1999.