— It is predicted that 70% of the chip area will be occupied by memories in future system-onchips. The minimization of on-chip memory hence becomes increasingly important for cost, performance and energy consumption. This paper proposes a novel memory size model for algorithms which map the variables of a system behavior to memories of a system architecture. To our knowledge, it is the first memory estimation approach that analyzes the variable lifetime for the system behavior, which consists of hierarchically-modelled and concurrently-executed processes and contains variables with different sizes. Experimental results show that significant improvements can be achieved.