This study examines simultaneously different methods of rating lithium response using the most widely used scale (the Alda Scale). Using continuous scores of lithium response or response categories leads to different predictors of response or non-response being identified. High B scale scores are associated with clinical features that are common in clinical populations such as the presence of alcohol or substance misuse or psychotic symptoms. This cross validation approach offers a useful methodology for exploring the utility of the Alda scale and other reliable measures of lithium response. A limitation of the findings is that rating scale is that it is primarily a retrospective measure of response.