This paper presents evidence that the quality of survey data on household incomes systematically improves across waves of a panel. Our estimates indicate that the effect of being interviewed for a second time is to increase the mean of reported monthly income by £142 (8 percent). Dependent interviewing - a recall device commonly used in panel surveys - takes effect only after a first interview. It explains approximately one third of the observed increase. The remaining share is attributed to changes in respondent reporting behaviour (panel conditioning). Our analysis suggests that falls in respondent confidentiality concerns are important in explaining the result.