This paper considers the information content of MLS descriptions and employs a significantly larger data set than previous studies. The analysis first catalogs the most frequently used terms by real estate agents in MLS descriptions. Using hedonic modeling, we estimate the effect of this qualitative information on transaction price and days on the market. Finally, we extend earlier empirical work by utilizing our larger MLS data set to forecast the probability that a house will sell after it is listed. This last contribution further sheds light on the role of qualitative information to infer property condition or circumstances surrounding the sale of the property.