This paper addresses the requirements for best practice in clinical assessment of people with cognitive deficits. It is proposed that a significant goal for assessment should be to enable formulation of a hypothesis regarding the areas of strength and weakness in functioning, in order to set appropriate aims for rehabilitation, and to evaluate the outcome of the rehabilitation process. Meeting this goal requires appropriate assessment before rehabilitation begins, during the rehabilitation process and after rehabilitation (by the clinician) has ended. Two approaches are contrasted: the ‘battery’ approach and the ‘hypothesis testing approach’. In order to measure change in both cognitive impairments and the handicaps they cause, testing is required that is reliable, sensitive and valid (measures changes that are of real consequence). It will be argued that assessment batteries often fail on all three of these requirements, and the hypothesis testing approach is superior in this regard. The paper provides an example of the use of the hypothesis testing approach in the assessment of word production impairments in acquired aphasia.