For wind energy conversion systems operating above the rated wind speed, the frequent pitch actions regulate the mechanical power as the rated one with the cost of blade and drive shaft loads. It is meaningful to maintain the desired power with appropriate pitch sensitivity related to the wind speed fluctuations, which can further reduce the mechanical loads of wind turbines with a longer service life. To quantify the blade pitch sensitivity, the blade pitch standard deviation is introduced to connect the pitch actions with the blade and drive shaft loads. Within a variable-weight model predictive control (MPC) strategy, both generator power output quality and load conditions are optimized through the pitch/torque participation coordination based on the Pareto analysis. Moreover, the MPC-weight matrix could be updated adaptively through the wind status assessment. The comparisons between the proposed strategy and the traditional gain scheduling PI one show the effectiveness. Several suggestions are also concluded for industrial wind turbines with MPC implementations.