The use of reflective learning journals to encourage higher order learning outcomes is a growing area in higher education research and practice. However, without a unified and clear definition of reflection, identifying and assessing reflection is problematic for educators. In an attempt to address this issue, in 1999 Kember and colleagues devised a coding scheme based on the work of Mezirow, to identify and assess levels of reflective thinking in students' written journals. We evaluated the usefulness of this coding scheme in a business education context. Findings revealed that the scheme was useful in identifying categories of reflective thinking. Inter-coder agreement was 0.802 which is satisfactory. On average, 65% of the journal content was coded as non-reflection and 35% as reflection. A further outcome of the research was to refine the coding scheme and to provide suggestions for its application in teaching practice. (Contains 1 table and 2 figures.)