Teachers' feedback on students' written work is an important aspect of pedagogy. However, theoretical views differ on what constitutes "good" feedback, both among applied linguists and academics in other disciplines. In-depth research needs to be carried out into the contextual difficulties of evaluating and assessing academic assignments, and the awarding of grades, especially by those who are relatively inexperienced in this work. This article reports aspects of a case study which explored the beliefs and practices of a group of untrained and inexperienced part-time tutors in a New Zealand university. Data were collected from a preliminary survey, individual interviews, "think aloud" and stimulated recall sessions, and focus group meetings. Extracts from the collected data are presented and discussed. The findings indicate that these tutors initially stated their belief that the purpose of providing feedback was to assist the students to improve their academic writing skills; however, it emerged that their primary concern was to justify the grades that they awarded. It is suggested that using a multi-method approach to data collection can bridge the gap between theoretical perspectives on what constitutes "good" feedback and what tutors actually believe and do in their everyday work.