The throughput performance of packet-based wireless networks can be modelled using Time-Stepped Simulation (TSS). In this type of simulation, instantaneous information related to transmission of the data can be obtained and send as feedback in resource allocation. This paper investigates the effects of time step size and load of the network on the reliability of throughput performance when TSS is used which is measured based on the consistency and accuracy of the results obtained. Results obtained show that a smaller time step requires longer simulation time before consistent results could be obtained. Results have also demonstrated that it is increasingly important to use a sufficiently small time step in cases with large arrival rate so as to avoid overestimation of throughput performance. For the mobile network defined herein, a time step size of 0.01 s is shown to be sufficiently small to produce reliable results in either cases of low or high user arrival rate.