The competitive response of co-existing species in a community to limited resources is an important determining factor responsible for community formation, maintenance, and succession. To reveal the plant competition mechanisms of community succession owing to habitat change, we studied the competition responses of co-existing species in old-field communities of the Loess Hilly Region under various water conditions. The experiment was carried out using a change of two factor, soil moisture, and planting density. There were two treatments of soil moisture, including invariant and changed water treatments. Invariant water treatment was carried out under severe (45% FC), moderate (60% FC), and mild (75% FC) water deficit levels, and changed water treatment was begun under mild water deficit levels (75% FC) and after non-water treatment for 50 days was changed to severe water deficit levels (45% FC). There were three levels of planting density determined by the Community Density Series (CDS) method (3, 12, and 21 individuals/m² in invariant water treatment; 5, 20, and 35 individuals/m² in changed water treatment). In this study, based on the CDS method, we measured and calculated proportional growth, relative biomass, relative proportional growth, and other indexes, to show the relative effects of competition and water stress, responses of relative competition ability, and competition hierarchies to changes of soil moisture, as well as the relationship between growth variation in water-welled period and mortality rate in severe water drought period. The results showed that (1) under different water and density conditions, the competition effects on the growth performance of most co-existing species in the community were greater than the drought effects. Community competition effects (4.30±0.49) based on the relative biomass had a higher value than the community drought effects (-23.56±20.38) and the accumulative effects (4.76±0.4) were greater than the sum of the two individual effects (-19.26±20.48), which means competition and water stress had synergistic effects on the community; (2) when the water conditions changed, the competitive hierarchy of community co-existing species was not exactly the same (P > 0.05) and corresponding changes occurred. Furthermore, the larger the difference of water conditions, the greater the difference in the competitive hierarchy of community co-existing species. The value of relative community competition intensity was highest (4.84±0.23) under mild water deficit levels, which partially supported the maximum growth rate theory, and indicated that higher levels of limited resources, resulted in more intense competition between community co-existing species; (3) for the changed water experiment, there was a positive correlation between individual size variation caused by the competition in the water-welled period and mortality rate in the drought period (P=0.015, r²=0.801). According to the two-stage hypothesis of resource dynamics, it showed that the community co-existing species competition was an important factor for the community formation in this experiment.