The issuance of weak patents is widely viewed as a fundamental problem in the current US patent system. Reasons that have been offered for the granting of weak patents by the US Patent and Trademark Office (USPTO) include examiners’ “rational ignorance” of the patentability of applications and pro-“customer” rules and institutions that create incentives for examiners to grant patents of dubious validity to their “customers”- applicants. In this paper, we study whether US examiners’ behavior in prior art search betrays their assessment of applications’ patentability. For a sample of US patents for which applications were also filed at the European Patent Office (EPO), we construct a measure of the fraction of prior art that is missed by US examiners. We find that this measure significantly explains the probability of receiving a patent at the EPO. The results are robust to different empirical specifications. US examiners’ prior art searches indicate that they are, on average, not “rationally ignorant”. On the contrary, they identify and dedicate more search effort to those applications that seem more problematic, because they bear the burden of proof of non-patentability. Our study offers empirical evidence that a systematic problem of weak patents likely exists, and suggests that the problem may be more strongly attributable to the pro-applicant rules and policies than to examiners’ ignorance. The current prevalence of weak patents does not appear to be caused at the margin by lack of resources at the USPTO.