AbstractEstimates of the level of unmet need for mental health treatment often rely on self‐reported use of mental health services. However, depressed persons may over‐report their use in relation to administrative records if they are highly distressed. This study seeks to replicate and explicate the finding that persons at a high level of distress report more mental health service use than recorded in their healthcare records. The study sample, N=36,892, 12 years and older, was drawn from the 1996/97 Ontario portion of the Canadian National Population Health Survey. Respondents were individually linked to their administrative mental healthcare records 12 months backward in time. Of these, 96.5% agreed to the link and 23,063 (62.5%) were linked. Almost two‐thirds of those who were depressed in the past year were currently at a high level of distress. Differential reporting of use for highly distressed persons in excess of 100% remained in the use of different types of physician providers after adjustments for other potential determinants of use. Telescoping was also not an explanation. The patterns of differential reporting between groups expected to diverge and converge in their recall ability were consistent with a recall bias. As this study was not able to rule out a recall bias, it further accentuates concerns about the impact of bias in the measurement of mental health‐service use and inferences made concerning the determinants of use. Copyright © 2004 Whurr Publishers Ltd.