We evaluate the stability of coefficient and willingness to pay (WTP) estimates for recreation services over two time periods. To address this question, we estimate a Random Utility Maximization (RUM) model of recreation demand, using two datasets from different time periods, but concerning the same study area. We then compare the estimation results and evaluate the temporal stability of preferences that drive recreation choices. The two datasets are on trips made by Delaware residents to beaches in the Mid-Atlantic region: Delaware, New Jersey, Maryland and Northern Virginia. The first dataset was collected using a mail survey in 1997 and the second dataset was gathered through an Internet survey in 2005. Besides the time periods, and the survey methods, there are also significant sample size differences between the two datasets. In the 1997 sample, 400 Delaware residents made at least one day trip, while in the 2005 dataset, only 50 Delawareans visited the beaches of interest.