A new procedure for estimating promotion, repetition and dropout rates for learners in South African schools is proposed. The procedure uses three different data sources: data from the South African General Household survey, data from the Education Management Information Systems, and data from yearly reports published by the Department of Basic Education. The data from the General Household survey are utilised to estimate repetition rates for learners in three different age groups. Keeping these repetition rates fixed, the data from the other two sources are used to estimate dropout and promotion rates, which are based on a birth-year-cohort approach for the different age groups. In particular, this procedure involves minimising the difference between actual flow-through rates and simulated flow-through rates for both the birth-year cohorts and age groups. The procedure gives different results when compared to published literature.