<p>Modelling results from PlioMIP2 (the Pliocene Model Intercomparison Project Phase 2) focussing on MIS KM5c; ~3.205Ma, suggest that global mean surface air temperature was 1.7 &#8211; 5.2 &#176;C higher than the preindustrial.&#160; This warming was amplified at the poles and over land.&#160; The results are in reasonable agreement with paleodata over the ocean.&#160; &#160;</p><p>Over the land the situation is more complicated.&#160; Model and data are in very good agreement at lower latitudes, however at high latitudes an initial data-model comparison shows much warmer mPWP temperatures from data than from models.&#160; &#160;</p><p>Here we consider possible reasons for this data-model discord at high latitudes.&#160; These include uncertainties in model boundary conditions (such as CO<sub>2 </sub>and orbital forcing), and whether there are local site-specific conditions which need to be accounted for.&#160; We also show that the seasonal cycle in mPWP temperatures at these high latitude sites has no modern analogue.&#160; This could lead to inaccuracies when comparing model derived mean annual temperatures with quantitative climatic estimates from palaeobotanical data using Nearest Living Relative methods.</p>