Calcium oxide can be an effective sorbent to separate CO2 at high temperatures. When coupled with a calcination step to produce pure CO2, the carbonation reaction is the basis for several high-temperature CO2 capture systems. The evolution with cycling of the capture capacity of CaO derived from natural limestones is experimentally investigated in this work. Long series of carbonation/calcination cycles (up to 500) varying different variables affecting sorbent capacity have been tested in a thermogravimetric apparatus. Calcination temperatures above T > 950 degrees C and very long calcination times accelerate the decay in sorption capacity, while other variables have a comparatively modest effect on the overall sorbent performance. A residual conversion of about 7-8% that remains constant after many hundreds of cycles and that seems insensitive to process conditions has been found. This residual conversion makes very attractive the carbonation/calcination cycle, by reducing (or even eliminating) sorbent purge rates in the system. A semiempirical equation has been proposed to describe sorbent conversion with the number of cycles based on these new long data series.