Based on a theoretical model of the mood-cognition interface, the prediction is derived and tested empirically that positive mood enhances constructive memory biases. After reading an ambiguous personality description, participants received a positive or negative mood treatment employing different films. Within each mood group, half of the participants were then questioned about the applicability of either desirable or undesirable personality traits to the target person. This questioning treatment was predicted to bias subsequent impression judgements in the evaluative direction of the (...) questioned attributes. As earlier research had shown that such a bias is stronger for negative than positive attributes (presumably because of the higher diagnosticity of negative attributes), a non-trivial test of the sup posed mood effect was possible. Positive mood should enhance constructive effects, but this should be most apparent for negative attributes. The empirical findings lend support to these predictions. Positive mood subjects' impression judgements were biased by the prior questioning even when they did not find the questioned attributes to be particularly applicable to the target person. By contrast, negative mood subjects' judgements were more consistent with their responses to the prior questioning and did not exhibit a constructive bias. (shrink)