Four 2-year trials were set up to test the effect of crop sequence and periods of bare fallowing on the conservation of moisture in soil profiles in a harsh, semi-arid environment. Amounts of moisture remaining after the harvest of the first-year crop were in the order: bare fallow > cowpea > sorghum. Subsequent losses over the crop-free winter period were generally low, ranging 0.07-0.17 mm day-1, according to the initial wetness of the profile and to the use or non-use of post-harvest tillage to control weeds and crop ratooning. Significant effects of the previous crop on the performance of the second-year test crop were seen in all four trials, with average yields in the proportion: ex-fallow, 1.0; ex-cowpea, 0.81; ex-sorghum, 0.40. Although soil moisture content at planting time was a major factor in these differences, other factors could be important according to the nature of the season. For example, residual nitrogen from the preceding cowpea crop was beneficial when the test crop received good rainfall. It is concluded that an alternation of short-cycle and long-cycle crops (e.g. cowpea/sorghum) with a short bare fallow in between uses water more efficiently than does a sequence, long bare fallow/sorghum, and it is also likely to be more acceptable to farmers.