Implicit measures such as the Implicit Association Test (OAT) have recently become popular as tools in research on evaluative conditioning. The reason is that these measures are thought to be impervious to changes in valence that are due to conscious propositional knowledge about the relation between the conditioned stimulus (CS) and the unconditioned stimulus (US). I present data that raise doubts about the validity of this assumption. In two experiments, attitudes towards CSs as measured by the IAT were in line with instructed CS-US contingencies even though the CSs and USs were never actually paired. The results therefore suggest that IAT effects can be biassed by conscious propositional knowledge.