Seismic risk analysis including soil – structure interaction (SSI) of harp type cable stayed bridge with support flexibility is presented which can be used for preliminary estimate of its probability of failure. The risk analysis procedure uses the format of probabilistic risk analysis (PRA) and considers the band limited white noise at the bedrock as the seismic input. The bridge deck is modeled as a beam supported on springs at different points. The coupled stiffness matrix of the springs is determined by a separate 2D static analysis of cable-tower-deck system in which flexibility of the tower base due to soil-structure interaction is included. Damping due to soil is incorporated by the equivalent modal energy method. The response of the bridge deck is obtained by the response spectrum method of analysis for multi-degree of freedom system. The PRA includes uncertainties of responses due to the variation in ground motion, material property, modeling and method of analysis, and uncertainties of the capacity due to the variation of ductility factor and damage concentration effect. Failure mode of the bridge is assumed to be bending failure of the bridge deck at the point of maximum bending moment. Probability of failure of the bridge deck is determined by First Order Second Moment (FOSM) theory of reliability analysis. A three span double plane cable stayed bridge is used as an illustrative example. The fragility curves for the bridge deck failure are obtained under a number of parametric variations. The parameters include, base flexibility, degree of correlation of ground motion, angle of incidence of earthquake, ratio of the components of ground motion. Study shows that flexible base condition provides significantly less value of probability of failure as compared to the fixed base. Further, angles of incidence, degree of correlation and ratio of components of ground motion have considerable effects on the probability of failure.