This experiment was conducted to evaluate five levels of nitrogen (Source: Urea) as 50, 75, 100, 125 and 150 kg ha^-1 and three split methods as T1=1:2-1:2 in pre-planting and late rosetting stages, T2=1:2-1:2 in pre-planting and early flowering stages, and T3=1:3-1:3-1:3 in pre-planting, late rosetting and early flowering stages of safflower. Treatments were arranged as factorial in completely randomized block design. Treatments were arranged as factorial in completely randomized block design. Interaction of N levels and method of application was not significant. Split application of lower amounts of nitrogen resulted in low fertilizer efficiency. So that grain yield by method of T3 was lower than those of methods T1 and T2. Thousand-grain weight and grains per head were affected by nitrogen application. Grain yield was increased by increasing the levels of nitrogen. Head per square meter was higher with three-split application than two-split application method . Based on these findings, nitrogen should be applied in split application respect to local conditions and phenological growth stages of safflower.