Many managerial practices and research findings are neither universal nor timerelevant. On one hand, their external validity is frequently low. They cannot be easily adapted from one organization to the other. On the other, even in the organizations in which they originate, the findings are often not durable. This is not surprising because practices and research findings are based on idiosyncratic or averaged situations which can't be easily found elsewhere. Furthermore, they are derived from past experiences and conditions which will be rarely the same in the future. A paradigmatic shift is necessary to reconcile apparently divergent approaches to management. The science of chaos pro- vides a new paradigm where two apparently irreconcil- able visions of management-rational and quasi-mech- anistic on one hand, unexpected and disorderly on the other hand-can be reconciled.