Traditional consumer protection law employs various disclosure requirements to respond to market imperfections that result when consumers are misinformed or unsophisticated. This regulation assumes that consumers can rationally act on the information that disclosure seeks to produce. Experimental results in psychology and behavioral economics question this rationality premise. The numerous reasoning defects consumers exhibit in these experiments would vitiate disclosure solutions if those defects also presented in markets. To assume that consumers behave as badly in markets as they do in the lab implies new regulatory responses. This Article sets out the novel and difficult challenges that such “regulating for rationality”—intervening to cure or to overcome cognitive error—poses for regulators. Much of the challenge exists because the contracting choices of rational and irrational consumers often are observationally equivalent: both consumer types prefer the same contracts. Hence, the regulator seldom can infer from contract terms themselves that reasoning errors produced those terms. Rather, the regulator needs a theory of cognitive function that would permit him to predict when actual consumers would make the mistakes that laboratory subjects make: that is, to know which fraction of observed contracts are the product of bias rather than rational choice. The difficulties exist because the psychologists lack such a theory. Hence, cognitive-based regulatory interventions often are poorly grounded. A particular concern is that consumers suffer from numerous biases, and not every consumer suffers from the same ones. Current theory cannot tell how these biases interact within the person and how markets aggregate differing biased consumer preferences. The Article then makes three further claims. First, regulating for rationality should be more evidence-based than regulating for traditional market imperfections: in the absence of a theory, the regulator needs to see what actual people do. Second, when the facts are unobtainable or ambiguous, regulators should assume that bias did not affect the consumer’s contracting choice because the assumption is autonomy preserving, administrable, and coherent. Third, disclosure regulation can ameliorate some reasoning errors. Hence, abandoning disclosure strategies in favor of substantive regulation sometimes would be premature.