For all mapping discrete-time models, characterized by a real parameter µ, Tustin’s model, corresponding to µ = 1/2, gives the smallest discretization error according to the result of order analysis. Using numerical simulations, this study shows that the value of µ, especially for closed-loop cases, that gives the smallest error is not always 1/2. It is also shown that, besides µ, the discretization accuracy is affected by the order of the plant and the bandwidth of the closed-loop control system. Also, some guidelines for selecting an appropriate value for µ are highlighted.