Using different datasets can be useful for comparing characteristics and outcomes across different populations or time periods. However, comparability of variables across datasets is often not straight forward. Ideally item sets would be harmonised across major studies. However, this is rarely the case. If exact duplication of wording in surveys was all that was required, very little cross survey comparison would be possible. On the other hand, having the same wording and answer categories may not be enough for two data sets to be fully comparable. For example sample characteristics or methodology may render comparability problematic. This presentation will use examples from four longitudinal surveys to explore some of the issues to consider when using multiple surveys for comparative purposes.