I build a model where potentially biased judges verify complex states by interpreting an imperfect signal whose noise captures factual ambiguities. In a sales and a financial transaction I show that judicial biases amplify and distort factual ambiguities, creating enforcement risk. To insure against such risk, parties write simple noncontingent contracts that optimally protect the party that is most vulnerable to judicial error. These results shed light on the empirical association between law and finance and rationalize salient features of real world enforcement regimes.