A series of five experiments was performed comparing abstract and thematic materials in Wason's (1966) selection task. The first was designed to test the effect of negated rule components on the facilitated performance usually found with thematic rules. No such facilitation transpired, however, and the succeeding four experiments investigated possible variables which could account for this result. None of the experiments showed any facilitation of performance on thematic rules and the pattern of responding with both abstract and thematic rules was consistent with ‘matching bias’ (Evans, 1972a). The results are discussed with reference to previous, apparently contrary, findings.