efficient multitasking systems based on Cortex-M3 architecture Modern micro-controllers provide enough processing power to benefit from the advantages of multitasking schedulers or operating systems even in the area of small, battery based or energy self-sustaining devices. Interrupt driven software designs provide maximum sleep times and energy savings in such devices. For a multitasking operating system, a software technique with minimal resource overhead must be found to transport data between interrupt service routines and tasks. This paper describes a queue implementation based on special low-level synchronization commands provided by the widespread Cortex-M3 micro- controller architecture. The proposed queue design provides a fast, low overhead data transport from task to task and between interrupt service routine and task. The implementation shows slightly lower latency (1%), higher bandwidth (18%) and has less memory overhead (33%) than the generic implementation provided by the well known FreeRTOS multitasking scheduler for embedded systems. It allows the use of multitasking schedulers even in ultra low energy embedded systems with few kilobytes of RAM.