Embodiments are described for prognostication of real time operating system (RTOS) health. An example computer-implemented method includes monitoring, for a task from a plurality of tasks being executed by the RTOS, an execution time, an inter-arrival time, and a blocking time. The method further includes computing an RTOS state of health value based on the execution time, the inter-arrival time, and the blocking time of each task from the plurality of tasks. The method further includes, in response to the RTOS state of health value being less than a predetermined threshold, initiating performance of an error handling.