The present investigation was conducted to study the effect of salt stress on sunflower growth through lysimeter experiments, statistical tools and model analysis. The experiment was conducted in Hetao Irrigation District, China. Three levels of soil salinity (EC1:5from 0.61-2.06 dS/m) were applied to 15 soil lysimeters and results indicated that both the average initial soil salinity and non-uniform salt distribution could affect relative yield of sunflower significantly and increasing initial salinity levels from 0.61-1.40 to 1.70-2.06 dS/m could reduce 32.7% relative yield. Furthermore, EC 0-10cm and EC0-40cm (before sowing), EC0.400cm (at seedling stage), EC20-30cm (at bud formation), EC 0-40cm (at flowering) and EC0-40cm (at milky ripe stage) were selected as salt predictor variables on the basis of multivariate regression analysis and established yield models by these indicators could achieve larger accuracy when using modified Jensen model (R=0.915). Moreover, the largest accuracy was obtained in the stepwise regression model established by agronomic and photosynthetic characteristics of sunflower in different growing periods (R=0.972).