BACKGROUND: Subsidiary findings in several studies indicate that the standard summary scores (total number of needs, met needs and unmet needs) of the Camberwell Assessment of Need (CAN) may conceal important differences among patient populations at the item level of the measure. The aim of this study was to investigate whether changes in need and need status at the item level are adequately reflected by changes in the summary scores. METHODS: In a longitudinal design assessments of need in 1997 and 2003 of 192 outpatients (mean age 45.4 years, 78.1% psychotic disorders) in routine mental health care were compared. RESULTS: None of the summary scores changed between 1997 and 2003. This result, however, was contradicted by significant changes in needs at the item level. CONCLUSIONS: The summary scores conceal changes in need on the underlying items and thus is recommended not to be used as dependent measures when comparisons among populations or between points in time are of interest.