Use of internet-mediated questionnaire assessment is growing in behavioural telehealth provision. Literature on web-based psychological testing indicates that--while tests used online should be validated for online use--such measures can be reliable and valid. However, use of normative data (especially data from paper-and-pencil samples) presents problems. Studies have shown score distributions may differ across testing media, even when comparisons are made between equivalent samples. Data from 3 different projects illustrate such (substantive) differences, and the errors that would result from using established norms to interpret internet-mediated assessments. Given representativeness issues with "internet norms", it is recommended that if online tests are used for clinical purposes, this should currently be done in a manner that does not rely on normative data.