The movement of many human interactions to the internet has led to massive volumes of text that contain high-value information about individual choices pertaining to risk and uncertainty. But unlocking these texts’ scientific value is challenging because online texts use slang and obfuscation, particularly so in areas of illicit behavior. Utilizing state-of-the-art techniques, we extract a range of variables from more than 30 million online ads for real-world sex over four years, data significantly larger than that previously developed. We establish prices in a common numeraire and study the correlates of pricing, focusing on risk. We show that there is a 15-19% price premium for services performed at a location of the buyer’s choosing (outcall). Examining how this premium varies across cities and service venues (i.e. incall vs. outcall) we show that most of the variation in prices is likely driven by supply-side decision making. We decompose the price premium into travel costs (75%) and the remainder that is strongly correlated with local violent crime risk. Finally, we show that sex workers demand compensating differentials for the risk that are on par with the very riskiest legal jobs; an hour spent with clients is valued at roughly $151 for incall services compared to an implied travel cost of $36/hour. These results show that offered prices in the online market for real-world sex are driven by the kinds of rational decisionmaking common to most pricing decisions and demonstrate the value of applying machine reading technologies to complex online text corpora.