To better understand and predict future climate change we often to look to past periods of global warmth as analogues and testbeds for models. The Miocene (23 to 5 million years ago) is probably the best analogue for the range of carbon dioxide concentrations, warming, sea level rise, and ice volume losses that we are appear to be largely committed to exploring in coming centuries. In my talk I give an introduction to the data we have for this time period and compare with climate model results. Emphasis will be placed on model failures and what we can learn from them. The main result is that climate models substantially underpredict the sensitivity of the climate system in the Miocene to the range of carbon dioxide concentrations (400-800ppm) that likely characterize our future.