One branch of the aid effectiveness literature (AEL) analyzes conditional models where aid effectiveness depends upon a conditioning variable z. The leading candidates for z are a good policy index and aid itself, so that the model has an aid squared term. In this paper, meta-analysis techniques are used (i) to determine whether the AEL has established the said interaction terms, and (ii) to identify some of the determinants of the differences in results between studies. We find no support for conditionality with respect to policy, nor with respect to aid itself. Copyright © 2009 John Wiley & Sons, Ltd.