Experimental studies and control descriptions of the human operator in recent years have extended the knowledge of adaptive behaviour in manual control from vague generalities about his versatility to quantitative descriptions. Considerable data have been gathered on the ability of the human to adapt to changes in input spectra, controlled element gain, polarity and dynamics, display modality, and the limits of controllability under a variety of situations. These results have indicated a greater adaptation versatility than expected from some earlier descriptions, but have also pointed out some of the restricted training conditions under which the rapid human adaptation may be expected to be demonstrated. When it comes to explaining how the human manages his remarkable adaptation abilities, less progress has been made. Improvements in the techniques for dynamic measurement of the adaptive process have been helpful in this regard but still fall short of what is required to observe the change in control law. Some appealing suggestions for detection models, model reference conitrol analogues and statistical decision theory pattern recognition algorithms have been put forth but are in no sense 'proven'. Primarily as a result of the measurement limitation and the lack of sufficient knowledge about adaptive control systems in general, the development of models for human adaptive control has been limited to general schema at this time.