Temporal stability of non-market values help to demonstrate the validity of value estimates and the reliability of benefit transfers, however evidence of value stability over time is limited. A literature review focused on meta-analysis studies shows that it is standard practice to adjust values for CPI changes between temporal periods, but that in about half of cases additional value increases over time were identified as significant. One implication is that benefit transfer may not be reliable in those cases. In this study temporal stability of recreation values for the Coorong in Australia has been assessed by travel cost model (TCM) and contingent valuation (CVM) experiments repeated seven years apart. The case study results identified temporal transfer errors were larger with the TCM (62%) than with the CVM (19%). Testing showed that the TCM models and values were significantly different over the seven year period, but the CVM models and values were not, even though the same recreation good was involved. These mixed results indicate that stated preference values may be less sensitive to temporal effects than revealed preference values.