The aims of this study was to electromagnetic spectrum (E.M.S) in determining the spectral signature of the soil according to their gypsum content at different moisture levels, to recasting of the gypsum using the best wavelength and to creation mathematic models for forecasting. The study was carried out at laboratories of college of agriculture, Anbar University. Soil with 42% gypsum content was brought from Salah Al-Den province and washed by distilled water to remove different amount of gypsum and finally to get soils with 3, 8, 17, 28 and 36% gypsum contents. Soils were packed in polyvinyl chloride tubes (PVC) with 7.5 cm diameter and 30 cm height with eight replicates for each gypsum level. Soils were moisten by capillary to maintain homogeneity of soil moisture along the column. Spectral reflectance of soil was measured by spectroradiometer at moisture level of 24%. The results showed an absorption band at 1750 nm wavelength was appeared. The depth of this band was increased as increasing of gypsum content in soil. There is highly significant correlation between gypsum content and spectral reflectance at each of infrared bands (B6) and (S.B). The correlation coefficients at the band B6 were 0.978, 0.979 and 0.972, respectively. Linear, Quadratic and cubic models were represent this relationship. Gypsum was correlated linear, quadratic and cubic by a positive regression relationship with B6. Linear, quadratic and cubic models were the best and more efficient to forecasting gypsum in soil with determination coefficient were 0.99, 0.99 and 0.99 respectively.