Cross-saturation effect in synchronous machines has been a subject of considerable attention in recent years. Inclusion of the cross-saturation in the machine model has two consequences. The first one, called here 'steady-state cross-saturation', consists of dependence of the steady-state d-q axis magnetizing inductances on the currents in both axes. The second one, termed here 'dynamic cross-saturation', is the existence of nonzero elements in the system matrix, that describe cross coupling between d-and q-axis. Dynamic cross-saturation appears in all the saturated machine models, regardless of the selected set of state-space variables, with the exception of the winding flux linkage state-space model. The aim of this paper is to compare behavior of various models when dynamic cross-saturation is neglected. It is shown that the impact of dynamic cross-saturation on accuracy depends on the selected set of state-space variables. In majority of cases omission of dynamic cross-saturation leads to very inaccurate results. However, it is found that for one particular class of models omission of dynamic cross-saturation has practically no impact on accuracy. These models therefore fully describe the complete saturation effect by means of only continuous variation of the d-q axis magnetizing inductances.