1. Passive greenhouse apparatus is commonly used to investigate the in situ biological response of terrestrial communities to global warming. 2. Although close conformity of greenhouse treatment effects to general circulation model (GCM) scenarios is widely claimed, no proof of such a relationship has yet been published. 3. Here, the relationship between passive greenhouse thermal environment and future climate conditions is considered using temperature data collected from within and without greenhouses deployed in the maritime Antarctic. It is revealed that in terms of thermal extremes, diel and annual variation, and overall distribution across the temperature spectrum, such apparatus achieves only poor simulation of GCM forecasts. 4. During summer, greenhouses induce an amplified daily range of temperatures, elevated maxima and accelerated rates of change. 5. During spring and autumn, diel temperature variation continues inside the greenhouses while snow cover protects the controls. 6. During winter, an inverse treatment effect occurs, in which the relative depth of snow cover causes lower temperatures in greenhouses than in controls. 7. These treatment effects differ significantly from GCM climate predictions. Changes recorded in the composition, structure and function of greenhouse biota may thus be artefacts of the methodology. 8. Thorough a priori testing of greenhouse treatment effects is recommended for future climate change studies that are to be conducted in environments subject to seasonal snowfall, solar elevation and day length.