We measure ambiguity attitudes for a representative sample of US households using a custom-designed module in the American Life Panel. Ambiguity attitudes vary substantially across people: half are ambiguity averse, 12% are ambiguity neutral, 37% are ambiguity seeking. Further, ambiguity attitudes depend on the likelihood of the ambiguous event: people tend to overweight low-likelihood ambiguous events and underweight high-likelihood events, a phenomenon called ambiguity-likelihood insensitivity. Consistent with theoretical predictions, higher ambiguity aversion is associated with less equity market participation, lower portfolio allocations to equities, and more retirement planning. High ambiguity-likelihood insensitivity is associated with a higher probability of being insured.