While the theory of the patent system is premised on the idea that patents will be used to exclude competitors, only a tiny fraction of patents are ever enforced. Legal and economic scholars have theorized as to how to identify valuable patents based on their individual characteristics. In this paper, we present the results of the largest empirical study ever conducted of the patent system. We compare the characteristics of litigated patents to those of issued patents generally, and we find important differences in a range of dimensions. These data confirm some predictions in the literature regarding patent value and refute others. New patents are more likely to be litigated than old patents. Foreign patent owners are less likely to litigate than domestic patent owners. Patents that issue to individuals or small companies are much more likely to be litigated than those that issue to big companies, though many of those patents have changed hands by the time they are brought to court. Patents that cite more prior art are more likely to be litigated, and those that are litigated tend to be cited more elsewhere. Most significantly, there are substantial differences between industries in the likelihood of patent litigation. Patents in the mechanical, computer, and medical device industries are significantly more likely to be litigated, for example, than patents in the chemical and semiconductor industries. In the paper, we explore the implications of these findings in detail. Taken together, the data give a profile of a few valuable patents that stand out from a field of ordinary ones. They are the patents that their owners spend the most time and money in prosecuting. They are the ones that competitors recognize as most important. They are concentrated in a few industries in which patents play a more significant role in encouraging innovation. And they are patents that issue to individuals or small companies with asymmetric stakes in patent litigation, not to large companies. These conclusions in turn have significant implications for the design of the patent system, patent reform efforts and patent valuation theories - implications we consider at the end of the paper.