Meta-analysis has been criticized for combining studies with different independent and dependent variables--the alleged "apples and oranges" problem. This study provides evidence for the inclusion of similarly valenced subscales from self-report mood scales in meta-analysis. Undergraduates (N=214) were randomly assigned to a "positive" or "negative" group to complete a checklist containing words from 12 different scales. Cluster analyses indicated that mood subscales formed a two-cluster positive and negative solution. All participants were assigned to the correct cluster. Discriminant analyses correctly classified more than 95% of participants for each mood scale. Results indicate that subscales from commonly used mood inventories can be combined in meta-analyses having positive or negative affect as the dependent variable.