Self tuning is an important new branch of control which is attracting increasing theoretical and practical interest. The objective of self tuning is to control systems with unknown constant or slowly varying parameters, so theoretical interest is concerned with the stability, performance and convergence of the recursive algorithms involved, while practical interest derives from its potential as a simple controller commissioning tool, both as a method for controlling time-varying or nonlinear plant over a range of operating points, and for dealing with batch problems where the plant or materials vary over successive batches. This paper summarises and expands previous work on the design of self-tuning controllers. It discusses the closed-loop properties of various classes of self tuner, convergence concepts and results, and some of the technical problems involved with implementing self tuners on small computers or microprocessors.