This paper studies what pooled binary-choice data reveal about stable individual preferences. Pooling choices across heterogeneous individuals creates an identification problem: it results in a distribution of shock-contaminated valuations. Thus, for example, the price at which the pooled choice probability equals one-half can be arbitrarily far from median WTP. With finitely many prices, we derive bounds for mean WTP and quantiles under restrictions on shocks and tails. Using data on the value of non-work time, the bounds are tight enough to exclude several pooled-logit and mixed-logit estimates. Changes in experimental design can lead to arbitrary changes in pooled-logit-implied WTP estimates.