Droughts are a pervasive natural hazard. According to a recent report of the Western Governors Association, an “extreme” or “severe” drought has been experienced in some part of the U.S. every year since 1895. Furthermore, among U.S. natural hazards, drought is the most costly -- NOAA's National Climatic Data Center estimates that since 1980, there have been 10 droughts among the 57 natural disasters costing over $1B, and that the average annual cost to the U.S. of these $1B or larger droughts has been almost $6B. The 1988 central U.S. drought alone, in 2002 dollars, cost almost $62B. Despite the economic importance of droughts, drought prediction methods are not well refined, and even characterization of droughts is more art than science. While methods for estimating the severity of droughts like the Palmer Drought Severity Index (PDSI) have been in existence for some time, their relationship to the physical factors by which drought is evidenced is questionable. Characterization of droughts is complicated by the fact that no single measure can fully describe them. Perhaps more importantly, even if one agrees to a drought definition that is potentially amenable to measurement (agricultural drought, for instance, arguably is related to abnormally low soil moisture) there are almost no long-term records of the relevant physical quantity (soil moisture). We describe a method by which agricultural and hydrological droughts can be characterized, using a method that we term severity-are-duration analysis (SAD), and use this method, and associated metrics, to evaluate the drought history of the continental U.S. from 1920 to present. We also show that over most of the U.S., droughts have become less lengthy and severe during this period, due primarily to generally wetter conditions. An exception is the Southwest and parts of the interior of the West, where, notwithstanding increased precipitation (and in some cases increased soil moisture and runoff), increased temperature has led to trends in drought characteristics that are mostly opposite to those for the rest of the country especially in the case of drought duration and severity, which have increased. Based on the land surface modeling methods that underlie our historic analyses, we describe a real-time drought monitoring tool for the continental U.S., and suggest methods by which the monitor might be extended to drought forecasting. We also comment on the role of climate feedbacks in drought prediction and predictability.