Real estate market data often contain outliers in the observations. Since outliers have a large influence on least squares estimates, robust regression methods have been recommended for this situation. Compares the performance of least squares and least median of squares, a robust method, in the estimation of price/income relationships for apartment buildings. Multiplicative models with multiplicative errors are estimated by means of natural log transformations. The study confirms the importance of employing robust methods for this application and implies this may well be so for real estate data sets more generally.