The paper will identify the characteristics of respondents that may be associated with the generation of inaccurate retrospective data and will investigate characteristics associated with a higher probability to give a wrong answer. It will do this by looking specifically at one employment question asked annually in the HILDA Survey which asks respondents to recall their employment status at the date of their last interview. This will be compared with the employment status of respondents, recorded with a high degree of accuracy during their previous interview, based on a series of questions which ask about their current employment status.