We present dynamical models of NGC 4494, which we built using our iterative method presented in a previous paper. These models are live N-body models consisting of equal-mass particles, and they are in steady state as confirmed by a fully self-consistent evolution. Our goals were of twofold. The first one – namely to test whether our iterative method could indeed be used to construct galactic models following given observational constraints, both photometric and kinematic – was fully achieved. Our method allowed us to go beyond a simple spherical model and to make full sets of rotating, axisymmetric models without any limitations to the velocity distribution. Our second goal was to understand the structure of NGC 4494 better, and more specifically to set constraints on its halo mass. For this we tried three families of models: without halo, with a light halo and with a heavy halo, respectively. Our models reproduce well the photometry and the kinematics, the latter except for specific regions where some non-equilibrium or non-axisymmetric structure could be present in the galaxy (e.g. the kinematically decoupled core). However, the lower-order moments of the velocity distribution (up to and including the second order) do not allow us to discriminate between the three haloes. On the other hand, when we extend the comparison to the higher-order moments of the velocity distribution obtained from the long-slit data, we find that our light halo model fits the data better than the no halo, or the heavy halo models. They also reproduce the shape of the angular dependence of the PNe velocity dispersion in the outermost parts of the galaxy, but not the amplitude of its azimuthal variation. This may imply that a more general class of models, such as triaxial, may be necessary for a better fit.