An increasing part of modern machinery, and virtually all electronic and telecommunications systems are software and hardware systems (SHS), the operation of which lies in the interaction between software (SW) and hardware (HW). High responsibility and complexity of modern SHS imposes strict requirements for evaluation and maintenance of the set reliability measures. In the paper, the influence of the choice of software reliability models on the assessment of reliability measures of software and hardware systems was investigated. To build the SHS reliability model, modeling technology of complex information systems based on Markov processes, in which SHS is represented as a discrete-continuous stochastic system was applied. As software reliability models that give the input value of the HW failure rate, the model with the HW complexity measure on the one hand, and the Goel-Okumoto models and S-shaped model as the most popular models on the other hand were chosen for the SHS reliability model. It is shown that using traditional HW reliability models, such as S-shaped and Goel-Okumoto leads to inflated SHS reliability measures, which makes it impossible to accurately assess the system operation risks. In addition, the behavior of the SHS readiness function, calculated based on the input data, received from the SHS reliability model with the complexity measure shows the non-monotonic dependence with the extreme point, and in this case this point is a point of minimum and is located in the area of small time values. Such differences in the SHS reliability estimates, made based on various HW reliability models, should be considered at stages of operation and routine maintenance of such systems.