Seedling biomass and allocation, transpiration water use efficiency (TWUE), and species competition between switchgrass (Panicum virgatum L.) and milkvetch (Astragalus adsurgens Pall.) were investigated in a pot-cultivated experiment under different levels of water availability. The experiment was conducted using a simple replacement design in which switchgrass and milkvetch were grown in growth chamber with ten seedlings per pot, in three combinations of the two species (0:10, 5:5 and 10:0). Five water treatments included sufficient water supply (HW), gradual soil drying from HW (DHW), moderate water stress (LW), gradual soil drying from LW (DLW), and re-establishment of LW conditions after 12 days of drying from LW (RLW). Water treatments were applied over a 15-day period. Biomass production and its partitioning, and TWUE were determined at the end of the experiment. Species competitive indices (competitive ratio (CR), aggressivity (A) and relative yield total (RYT)) were calculated from the biomass dry weight data for shoots, roots and total biomass. Water stress significantly reduced seedling biomass production but increased root:shoot ratios in both monocultures and mixtures. In the RLW treatment, only switchgrass monocultures displayed compensatory biomass production and TWUE, while both species demonstrated compensatory growth in the mixture. Switchgrass was the dominant species and much more aggressive than milkvetch in the LW treatment, while in the other four treatments milkvetch was the dominant species as measured by the positive value of aggressivity and higher values of CR. The total biomass RYT values of the two species were higher than 1.0, indicating some degree of resource complimentarity. In the two-species mixture, although the biomass production was lower than that of milkvetch in the monoculture, there was better TWUE, especially under low and fluctuating water availability.