Abstract: The objective of this study was to evaluate existing constitutive models currently used by State Department of Transportations (DOTs) and to develop an improved model for predicting resilient modulus (Mr) of cohesive soils from simple soil properties typically measured in DOT laboratories in preference to expensive and complex Mr laboratory testing. The data set used consisted of cohesive soils typical of those used for subgrades in Ohio. Thirteen representative cohesive soils representing A-4, A-6, and A-7-6 soil types collected from road construction sites across Ohio, were tested in the laboratory to determine their basic engineering properties. Mr tests were conducted at three different moisture contents (dry of optimum moisture content, optimum moisture content, and wet of optimum moisture content). Additional tests were performed on samples compacted to optimum conditions but allowed to fully saturate. Mr predicted from six existing models studied showed wide scatter and poor correlation with the measured Mr. An improved constitutive model was developed to account for the effects on Mr of the stress state of the soil and its engineering properties obtained from simple laboratory tests. While most of the existing models investigated in this study significantly overestimated the Mr of a cohesive soil, the proposed model predictions are close to the experimental values and are in most cases a slight underestimation. This implies that Mr values predicted by the proposed model are generally slightly conservative, and can be safely used in the design of flexible pavements to be built on cohesive soils. The proposed model can be a useful and reliable tool for estimating Mr of cohesive subgrade soils using basic soil properties and the stress state of the soil.