Surgical needle tracking is an important element of high level automated operations using surgical robotic system. However, in conventional algorithms, the needle has the singular posture which the system cannot be detected. This paper discusses a robust, efficient tracking algorithm capable of estimating all positions and postures of the needle in an image space. The state parameters of needle are estimated using the projective contour models of a 3-D CAD model of the needle. The likelihood of the contour model is measured using the distance transformation, and the particle filter estimates the state of the needle. Experimental results indicate that our proposed method is accurate.