This article focuses on the challenges involved in conducting evaluations of interactive radio programmes in South Africa with large numbers of schools, teachers, and learners. It focuses on the role such large-scale evaluation has played during the South African radio learning programme's development stage, as well as during its subsequent sustained implementation phase. The model evolved for evaluation of interactive radio instruction is based within the context of a shoestring/real-world evaluation tradition, where funding for internal evaluation has been limited over the period of the programme's development to scale, necessitating focused use of resources in a longitudinal evaluation design. The evaluation approach is participatory and multimethod, linking the requirements of external summative evaluation conducted for accountability, with data yielded by internal (formative and developmental) evaluation. This is done through internal and formative evaluative studies of limited scope, combined with developmental classroom-based evaluation based on the logic of project team members working with teachers to promote the programme's self-evaluative capacity. (Contains 3 tables.)