The definition of desertification accepted in the ad hoc conference held by UNEP in Nairobi in 1977 and confirmed at the Earth Summit on Environment and Development held in Rio de Janeiro in 1992 is: ‘arid, semi-arid and dry-subhumid land degradation’. There is no global long-term trend in any rainfall change over the period of instrumental record (c. 150 years), but there has been an increase of 0·5°C in global temperature over the past 100 years. This increase seems partly due to urbanization, as there is no evidence of it resulting from atmospheric pollution by CO2and other warming gases (SO2, NO2, CH4, CFH etc.). On the other hand, the thermal increase is uneven, increasing with latitudes above 40° N and S. The increase is only slight or non-existent in subtropical and inter-tropical latitudes where most arid and semi-arid lands lie. This, incidently, is consistent with Global Circulation Models (GCM) — derived scenarios. The study of tree-rings, lake level fluctuations and pollen analysis confirm the existence of climatic fluctuations, but with no long-term trends over the past 2000 years. A possible increase of 1–3°C in arid lands over the next 50 years due to a doubling of the CO2content of the lower atmosphere to 700 p.p.m., as assumed by most scenarios stemming from GCM, would increase global potential evapo-transpiration (PET) by some 75–225 mm year-1. The ratio of mean annual precipitation to PET would then decrease by about 4–5%, assuming that no substantial changes in rainfall took place in arid and semi-arid lands. However, the impact of CO2on plants would boost photosynthesis and, therefore, primary productivity; it would also increase water-use efficiency via the reduction of stomatal conductance. It is therefore at present difficult to predict the net balance of these two opposite consequences or to prophesy which phenomenon would prevail: increased aridity or higher productivity and more efficient water use. At all events, the possible effect of a climatic fluctuation (or change) of the magnitude envisaged would have a trivial consequence on arid environments, as compared with the past and present impact of humans and their livestock. Drought has always been a normal recurrent event in arid and semi-arid lands. Strategies and tactics to mitigate its consequences via improved land-use and management practices are analysed.