We estimate changes in the total recreation value over a 20 year time period of a large newly established forest, using a mixed specification of random utility models and a geographic information system. The models are estimated using data from two identical surveys in 1977 and 1997. We conduct three different spatial value transfers and test these on the new forest. Results suggest that the recreation value of the new forest increased 70 times over the 20 years, primarily due the maturing of the forest and changed patterns of visitor behaviour. The value transfer to the new forest ranges between an underestimate of 57% to an overestimate of 349%, depending on the sampling of the choice set used as the study sites in the transfer.