Studies on anchoring effects indicate that judgments can be biased by previous comparisons to high- or low-anchor values. Anchoring effects have been demonstrated in many domains and they have been found both for relevant anchors that provide partially valid information concerning the assessed target as well as for irrelevant anchors that clearly don't. Based on previous findings it has been argued that anchoring effects are independent of the relevance of the anchor. In research on multiple-cue inferences it has, however, been found that individuals are highly sensitive to the relevance (validity) of cues. In two studies on sentencing decisions we show that relevant anchors influence sentencing decisions to a larger degree than irrelevant ones. We consistently find an effect of relevance for high anchors. Results still remain a bit mixed since the effect of relevance did not hold for low anchors that were introduced in the second study.