The present research study was carried out during the years 2009-2010 and 2010-2011 to identify the relationship between carbon isotope discrimination (δ) of grain and leaf, yield of winter wheat, and irrigation water productivity (IWP) under different water conditions in a semiarid climate. The field experiments were conducted with four different irrigation treatments (I1: rainfed: I2: irrigate when calculated soil water depletion is 60 mm below field capacity (full irrigation); I3: two irrigations maximum, one at tillering and another at grain filling; I4: no irrigation after establishment until heading, after which irrigate when soil water depletion is 60 mm below field capacity). The leaf (δL) and the grain (δG) carbon isotope discriminations, biomass, and grain yield (GY) were measured in the experiments and the harvest index (HI) and IWP were calculated. At the end of the study, taking a 2-year average, GY and HI were found to be 3.35 t ha-1, 4.53 t ha-1, 4.13 t ha-1, and 4.37 t ha-1 and 29%, 31%, 31%, and 32%, respectively, according to the treatments. The results showed a significant positive linear correlation between δ and GY. These results highlight that grain δ at the pre-anthesis stage could be beneficial for predicting yield under well-irrigated conditions. The highest IWP value was obtained from I4 treatment. δ-IWP and GY-IWP were negatively correlated. IWP can be indicated as an advantage in deciding about limited irrigation regimes for wheat production in arid areas. The results of the present study show that full irrigation treatment (I2) could be recommended in areas with no water shortage conditions. Moreover, limited irrigation such as treatment I4 at 4.8% level produced only optimum yield reduction and had the potential for saving approximately 50% of irrigation water.