An analysis is performed on the transfer test data from an experiment which factorially combines percent reward (50% vs 100%) and duration of confinement (2 vs 20 sec) in the presence of cue values from two redundant stimulus dimensions. Rats perform better on test trials which pair the previously negative cues with novel compound cues than on trials in which the previously positive cues are paired with a novel compound. Long confinement durations and partial reward during acquisition increase this difference. The results give further evidence that novelty affects test performance in single-dimensional transfer tests following redundant cue training.