The theory of hedonic models states that it is possible to precisely describe the price of a heterogeneous commodity by a set of its characteristics. However, the quality of the obtained results depends on the completeness of the set of significant attributes of the commodity used for estimation, as well as the statistical correctness of the model. In cases where the set of explanatory variables is numerous, and when the problem of multicollinearity occurs, it is not possible to use standard methods of estimation (such as OLS). The aim of this article is to examine the usefulness of the partial least squares method (PLS) in the estimation of hedonic models with a large number of correlated characteristics. It is shown that the use of PLS can yield better results than the alternative solution - the removal of problematic variables from the data set. Empirical research was carried out for selected groups of durable commodities. The databases were created using a tool for collecting data from web pages developed by the author.