Houses are expensive illiquid long-lived assets traded in an imperfect market. The imperfect nature of the market is a result of several factors, including the heterogeneous nature of the houses, asymmetric information among buyers and sellers, search costs, and transaction costs. These aspects of housing markets make the selling transaction a process rather than an event and, therefore, we observe two outcomes of this process: selling price and liquidity. There is evidence that the degree of liquidity of housing assets changes over time. That is, adjustments in the housing market seem to take place not only through prices and quantities but also through the degree of liquidity; i.e., how easy or difficult it is to sell a property as of a specific date. While much analysis has been done on how various attributes influence the selling price of a house, less work has been done on the factors influencing liquidity. This paper uses quantile regression for duration analysis allowing for a flexible specification of the functional relationship and of the error distribution. Censored quantile regression addresses the issue of right censoring of the response variable that is common in duration analysis. We apply this method to a large sample of housing transactions in Finland as well as to different subgroups of housing sales and sales during different housing market cycles. This way we examine how the effect of attributes on marketing duration may differ along the distribution of marketing times.