A review of the literature has highlighted that there are many approximations and ambiguities in practices for assessing Safety Integrity Level (SIL) of Safety Instrumented Systems (SIS). The aim of this paper is to make at first a review of mathematical inconsistencies in the practices for assessing SIL. Secondly, it introduces an unambiguous approach for risk reduction based on a discrete time approach. Thirdly, it proposes a predictive model to quantify the average number of failures that a SIS will experience during its life cycle. Finally, to the ends of validation, a comparison is made between the results obtained by applying the proposed novel approach, the formulas given in the IEC61508 standard and Monte Carlo simulations.