This paper proposes an adaptive kanban and production capacity control system as a new production planning and control mechanism for Just-in-Time environments to minimize the long term average inventories, Work-In-Process, backorders and operating costs. It is based on the adaptive kanban system proposed by Tardif and Maaseidvaag (2001), but dynamically adjusts both the number of kanbans and the level of production capacity with respect to inventories and backorders. It is expected to be more resistant to changes in the demand than previous pull ordering mechanisms. We present how to evaluate its performance for the case of a single-stage, single-product manufacturing process with exponential processing times and demand following a Poisson process. Simulation results under variable demand means are presented.