There is good evidence that higher global temperatures will promote a rise of greenhouse gas levels, implying a positive feedback which will increase the effect of anthropogenic emissions on global temperatures. However, the magnitude of this effect predicted by the available models remains highly uncertain, due to the accumulation of uncertainties in the processes thought to be involved. Here we present an alternative way of estimating the magnitude of the feedback effect based on reconstructed past changes. Linking this information with the mid-range Intergovernmental Panel on Climate Change estimation of the greenhouse gas effect on temperature we suggest that the feedback of global temperature on atmospheric CO₂ will promote warming by an extra 15–78% on a century-scale. This estimate may be conservative as we did not account for synergistic effects of likely temperature moderated increase in other greenhouse gases. Our semi-empirical approach independently supports process based simulations suggesting that feedback may cause a considerable boost in warming.