This thesis is concerned with the development and application of feedback control techniques for active sound control. Both fixed and adaptive controllers are considered. The controller design problem for active sound control is formulated as a constrained optimisation problem with an H₂ performance objective, of minimising the variance of the control error, and H₂ and H∞ design constraints involving control power output, disturbance enhancement, and robust stability. An Internal Model Controller with an FIR control filter is assumed. Conventional H₂ design models for feedback controllers are studied first. Although such controllers can satisfy the design constraints by employing effort terms in the quadratic cost function, they do not achieve the best possible performance, and when adapted using LMS-based algorithms, they suffer from instabilities if the plant response varies significantly. Improved H₂/H∞ design methods for fixed and adaptive controllers are then developed, which achieve the best H₂ performance under the design constraints, offer an improved stability when made adaptive, and in general outperform the conventional H₂ controllers. The H₂/H∞ design problems employ convex programming to ensure a unique solution. The Sequential Quadratic Programming method is used for the off-line design of fixed controllers, and penalty and barrier function methods, together with frequency domain LMS-based algorithms are employed in the H₂/H∞ adaptive controllers. The controllers studied and developed here were applied to three active sound control systems: a noise-reducing headset, an active headrest, and a sound radiating panel. The emphasis was put on developing control strategies that improve system performance. First, a high performance controller for the noise-reducing headset was implemented in real-time, which combines analogue and adaptive digital controllers, and can thus reject disturbances which has both broad-band and periodic components. Then, robust H₂/H∞ controllers were designed for an active headrest system using a virtual microphone arrangement to reject the sound at the listener's ears.