Optical jitter, the centroid-shifting of a light image, concerns engineers and scientists working with lasers and electro-optical systems. Even micron-level relative motion between individual optical components such as mirrors and lenses causes optical jitter, resulting in pointing inaccuracy, blurred high-resolution images, and poor nanotechnology quality. Typical jitter control technology uses fast-steering mirrors to correct for structural and acoustic disturbances in the beam train. Unknown or time-varying disturbance characteristics necessitate a controller that can adapt its parameters in realtime. The application of one such adaptive feedback controller algorithm has been proposed by the authors. The algorithm uses a technique known as Q-parameterization to structure the controller as a function of plant coprime factors and a free parameter, Q. An inherent property of this structure is the formation of a disturbance estimate based on subtraction of the controller influence from the feedback signal. The free parameter, Q, filters this estimate to form a portion of the control signal. If the controller influence on the feedback signal is estimated from accurately modeled plant dynamics, the disturbance estimate contains no feedback information allowing Q to be designed in an open-loop fashion. A gradient descent Least Mean Squares (LMS) algorithm updates the coefficients of the filter Q in realtime to minimize the frequency-weighted RMS jitter. Experiments on an optical jitter control testbed with Q set to a 200-tap digital finite impulse response (FIR) filter resulted in jitter reductions of 35% - 50%, without requiring prior knowledge of the disturbance spectrum.