This paper proposes a velocity control approach for the light electric bike with human power and disturbance estimation. The implementation of the disturbance observer in both the open-loop and closed-loop is tuned in terms of the dc gain of a Q-filter. The torque control law assumes that the slow varying part of disturbance has been successfully estimated and compensated, and is obtained from optimal control method to achieve multi-objective performances regarding the external disturbance input, control signal magnitude, and velocity tracking error. The optimal control signal is derived in terms of linear matrix inequality conditions. Based on the parameters and specifications of the EL-168 electric bike produced by KENTFA Advanced Technology, Taiwan, the design results are verified through time-response simulations.