Nowadays, in industrial control applications, is rather usual to sample and update different variables at different rates, although it is common to consider all these activities equally and regularly spaced on time. These applications are implemented on real-time operating systems by decomposing them into several tasks in such a way that pre-emption and blocking may appear due to task priorities and resource sharing. This could imply the presence of delays, leading to a non-regular periodic behaviour and, as a result, the control performance can be degraded. In order to undertake this problem, a solution based on a modelling methodology for non-conventional sampled-data systems is proposed. This technique permits the consideration of any cyclic sampling pattern. Thus, these delays can be considered in the modelling step, and later on, a non-conventional controller based on this model can be designed. In this way, if the considered non-conventional control system is implemented assuming a real-time operating system (Tornado-VxWorks, in this case), a clear performance improvement can be observed.