Purpose – Previous research on going concern warnings has focused mainly on prediction accuracy, giving little attention to going concern warning ambiguity or treating the issue of how auditors phrase their going concern warnings in a cursory manner. This paper argues that these issues are interrelated and need to be considered jointly. The purpose of the present paper therefore is to measure auditors’ prediction accuracy, to determine how common going concern warning ambiguity is, to explore the phrasing of going concern warnings and to highlight any patterns found in this analysis. Design/methodology/approach – The paper is based on empirical evidence drawn from the most recent annual and audit reports of all Swedish limited companies’ that went bankrupt in 2010. The audit report sample consists of 4,718 companies, and the going concern warning sample consists of 797 companies.Findings –Consistently with previous research, the findings demonstrate that Swedish auditors issue going concern warnings only in extremis and that it is not uncommon for the going concern warnings to be phrased ambiguously. The findings also reveal that going concern warnings can be categorized as clear, medium or fuzzy with several subordinate variations within those main patterns. Originality/value – Within the relatively unexplored field of going concern warning ambiguity, this study is based on a larger sample than that of the few existing previous studies. The findings provide a fine-grained understanding of the complex phenomenon of wording ambiguity, showing that in an audit context where the phrasing of going concern warnings is elective, where most firms are small and medium-sized and where auditors are required to assess if shareholders’ equity is spent partially or in full, ambiguous going concern warnings are both frequent and varied.Key words Audit report, accuracy, bankruptcy, going concern warnings, phrasing patterns, wording ambiguity, SwedenPaper type: Research paper