This study explores the challenges of assessing student workplace performance during work-integrated learning. It highlights the need for, yet difficulties with, combining positivist and constructivist assessments where workplace supervisors make evaluative judgements on performance yet students are also agents in their own assessment. It examines the ratings awarded by 163 workplace supervisors for 213 business undergraduates completing a work placement as part of their degree program in Western Australia. Students were rated on 17 capabilities associated with employability and results indicate, in alignment with previous studies, a tendency among supervisors to assign inflated marks across capabilities. The mean capability rating awarded to each student was significantly higher than their weighted course average, suggesting workplace supervisors mark more highly than academics in coursework units. To identify solutions to manage leniency bias, the study examined variations in supervisor ratings for a range of personal and contextual variables such as gender, organisation size, work area and sector. Although supervisor ratings were inflated, they were consistent across the sample with variations recorded for only four capabilities in certain work areas. Reasons for leniency bias among workplace supervisors are explored in light of the findings and alternative approaches to evaluating student workplace performance are presented.