From a normative perspective the order in which evidence is presented should not bias legal judgment. Yet psychological research on how individuals process conflicting evidence suggests that order could matter. The evidence shows that decision-makers dissolve ambiguity by forging coherence. This process could lead to a primacy effect: initial tentative interpretations bias the view on later conflicting evidence. Or the process could result in a recency effect: the evidence presented last casts decisive light on the case. In two studies (N1 = 221, N2 = 332) we test these competing hypotheses in a mock legal case. Legal orders sometimes even expect judges to provisionally assess the evidence. At least they have a hard time preventing this from happening. To test whether this creates or exacerbates bias, in the second dimensions, we explicitly demand experimental participants to express their leaning, after having seen half of the evidence. We consistently observe recency effects and no interactions with leanings. If the legal order wants to preempt false convictions, defendant should have the last word.