Econometric hedonic models encounterseveral theoretical and practicaldifficulties when applied to the realestate market, such as downwardbiases in the estimation of hedonicprices, subjective decisions in themeasurement process of categoricalattributes, frontier problems related toan imperfect information frameworkand uniequational specification. Manyof these are linked to the parametricapproach.Artificial Neural Networks (ANN)provide an attractive alternative: betterdwelling prices estimates, avoidanceof bias at different market segments,direct use of categorical data and fulluse of the information available. Theprice to be paid is the difficulties inthe economic interpretation of networkparameters. Nowadays, if the finalobjective to produce better estimates ofthe transaction prices, this methodologyshow lower errors, provided of a broadrepresentative database of sales arerecorded. A case study is presentedfor a medium size city in the South ofSpain.Keywords: Urban Economics; Hedonic Models; Neural Networks.