BACKGROUND: Lost productivity from attending work when unwell, or "presenteeism", is a largely hidden cost of mental disorders in the workplace. Sensitive measures are needed for clinical and policy applications, however there is no consensus on the optimal self-report measure to use. This paper examines the sensitivity of four alternative measures of presenteeism to depression and anxiety in an Australian employed cohort. METHODS: A prospective single-group study in ten call centres examined the association of presenteeism (presenteeism days, inefficiency days, Work Limitations Questionnaire, Stanford Presenteeism Scale) with Patient Health Questionnaire depression and anxiety syndromes. RESULTS: At baseline, all presenteeism measures were sensitive to differences between those with (N=69) and without (N=363) depression/anxiety. Only the Work Limitations Questionnaire consistently showed worse productivity as depression severity increased, and sensitivity to remission and onset of depression/anxiety over the 6-month follow-up (N=231). There was some evidence of individual depressive symptoms having a differential association with different types of job demands. LIMITATIONS: The study findings may not generalise to other occupational settings with different job demands. We were unable to compare responders with non-responders at baseline due to anonymity. CONCLUSIONS: In this community sample the Work Limitations Questionnaire offered additional sensitivity to depression severity, change over time, and individual symptoms. The comprehensive assessment of work performance offers significant advantages in demonstrating both the individual and economic burden of common mental disorders, and the potential gains from early intervention and treatment.