Recent studies have demonstrated that cognitive conflict, as experienced during incongruent Stroop-trials, is automatically evaluated as negative in line with theories emphasizing the aversive nature of conflict. However, while this is well-replicated when people only see the conflict stimuli, results are mixed when participants also respond to stimuli before evaluating them. Potentially, the positive surprise people feel when overcoming the conflict, allows them to evaluate the experience as more positive. In this study, we investigated whether task experience can account for contradictory findings in the literature. Across three experiments, we observed that responding to incongruent stimuli was evaluated as negative on the first trials, but this effect disappeared after 32 trials. This contrasted with the results of a fourth experiment showing that the negative evaluation of incongruent trials did not disappear, when participants could not respond to the conflict. Finally, a reanalysis of three older experiments corroborated these results by showing that a positive evaluation of conflict only occurred after participants had some experience with the task. These results fit with the idea that we need to create outcome expectancies (lower expectancies for being correct on incongruent trials) before we can experience the resolution of conflict as positive.