In this article authenticity is defined by componential concepts, common to spoken and written communication through language, which frame assessments of interchanges produced for "high-stakes" examination purposes. They are referents for analysis and critique of International Baccalaureate (IB) Diploma Programme evaluations of second and foreign language performance in international contexts. Experimental data derived from wide-ranging, criterion-referenced measurements triangulate results for IB-derived assessments. Within this system, potential improvements to practice in evaluating interactive language use are explored. Through focussing on communicative authenticity, validity and reliability problems in grading students' language proficiency are investigated. Greater consistency in test-task and criteria design for qualitative assessment of language use, it is argued, better matches production samples with programme aims and objectives. The overall credibility of typical evaluation systems may thereby be enhanced. (Contains 5 figures.)