For the development of more sustainable farming systems, two conservation tillage field experiments were conducted on the western Loess Plateau at Dingxi and Xifeng, Gansu, China from 2001-2005. Dingxi had a lower rainfall (400 mm) with one crop a year, whereas Xifeng had higher rainfall (550 mm) with 3 crops in two years. The crop rotation was spring wheat-field pea at the Dingxi site and corn-winter wheat-soybean at the Xifeng site. The basic design was a 2 x 2 factorial design (tillage x straw cover) with 4 replicates for both experiments with plastic cover as additional treatments at the Dingxi site. Both experiments were designed as fully phased experiments that each crop presented at least once in each year and two years as one cycle. At the Dingxi site, results showed that no-till with straw cover treatment had the highest grain yield for both spring wheat (2.4 t/ha) and field pea (1.8 t/ha), which were 16% and 15% higher than those on the conventional practice with repeated ploughing and stubble removal, respectively. However, no-till without stubble cover had the lowest grain yield. This highlights the importance of stubble retention in contributing to crop yield increases and the deleterious effect on crop yield of adoption no-till alone. Plastic film cover had risk of environmental problem although it could increase grain yield in short term. At the Xifeng site, there were no yield advantage for no-till practice, but stubble cover increased grain yield significantly for soybean (P < 0.01), but not for corn and winter wheat. The labour and energy savings from reduction of tillage operations could be attractive for the adoption of conservation tillage. In addition, the long-term effects of conservation tillage on soil physical and chemical properties would be beneficial to the farmers in managing these highly erodible soils.