The mid-Holocene period (from approximately 9000 to 6000 years before present) is often suggested as an analogue for enhanced greenhouse warming. The changes in net radiative forcing at the top of the atmosphere are very different; increases in greenhouse gases producing a small annual mean warming of little seasonal or latitudinal variation, whereas during the Holocene the annual mean did not change but there were large seasonal and latitudinal variations. Two climate model experiments, one in which CO2 amounts are doubled and the other in which the value of the earth's orbital parameters are altered to those appropriate to 9000 years before present (BP), are compared. Any similarity in the simulated response is found to be limited to the northern continents and, even them, the mechanisms producing the changes differ between the two experiments. Assuming that the gross behavior of the model is realistic, the Holocene is not a good analogue for a “greenhouse” warming. Furthermore, as the mechanisms operating in the two experiments are different, a model which produces a realistic simulation for the mid-Holocene and present climate need not necessarily produce a reliable simulation of greenhouse warming. However, a comparison of simulated climates for the mid-Holocene and that reconstructed from paleoclimatic data may help to constrain the existing range of subgridscale parametrizations used in climate models.