Abstract: A sensitivities analysis based emergency load shedding optimization method for the HVDC receiving end system was proposed in this paper. Firstly, the dynamic responses when HVDC lines were blocked were analyzed. The transient stability margins of rotor angle, frequency and voltages were defined quantitatively. The sensitivities between the amount of load shedding and transient stability margins were calculated. Then the emergency load shedding optimization problem was modeled taking the economic costs of load shedding as the objective function and the limits of transient stability margins and load shedding ratios as constraints. The artificial intelligence algorithms were hard to apply in practice due to heavy computation burdens. To solve this problem, the nonlinear dynamical optimization problem was approximated with a linear programming problem according to the sensitivities. It was solved iteratively until a convergence was reached. The multi-point start technology was used to avoid local optimization solutions. Finally, this algorithm was applied in Henan Province power grid to validate its effectiveness. The results show that it converges rapidly and can reduce economic cost of control efficiently while the transient stability of the receiving end system was guaranteed.