This research is concerned with the question of performance in scientific research, and focusses on the potential of Adaption-Innovation theory (Kirton, 1976) for providing insights into individual innovativeness. Using empirical data from four large research organisations, a taxonomy of scientists is developed using the Kirton Adaption-Innovation (KAI) inventory. This taxonomy breaks new ground in its use of the sub-scales of KAI. It shows that the use of the total KAI scale, as in previous research in the literature, is a conflation which conceals important insights. The research also breaks new ground in its conceptualisation of research performance. Two dimensions of performance are hypothesised: creative performance and skills performance. The evidence suggests that the distinction is meaningful and that the two dimensions are essentially orthogonal. The taxonomy developed in this research identifies four types of scientists according to their location on the '0' and 'E' sub-scales of the KAI. It is in the distinction between two types possessing similar mid-range KAI scores that the research makes a notable contribution to the literature. These two types are shown to be very different in terms of their performance, job satisfaction and other characteristics. It is through these insights that the research offers the prospect of an instrument of value in the deployment of research scientists. Finally, concerns about the conceptual status of the KAI are developed. The KAI is critically reviewed, and the evidence presented seriously challenges the claim that the KAI is purely a measure of cognitive style. Criticism is focussed on the 0 sub-scale which, it is argued, contains items measuring level of cognitive ability. A refined KAI is developed and evaluated using a sample of post-graduate students of management. It is demonstrated that sub-scales can be derived which are more homogeneous conceptually and give nearly orthogonal measures.