High-frequency seismograms mainly consist of incoherently scattered waves. Although their phases are more or less random, their envelopes show smooth and stable variations depending on frequency and distance. Envelope modeling can thus be used to infer stochastic parameters of the heterogeneous Earth medium. Radiative transfer theory (RTT) describes energy transport through a random heterogeneous medium neglecting phase information and has been frequently used to simulate observed mean square (MS) envelopes of high-frequency waves. The radiative transfer equations can be numerically solved by Monte Carlo simulations. So far, mostly isotropic scattering and acoustic approximations have been used. Here we present an extension of the Monte Carlo method to the full elastic case including P, S, and conversion scattering where the single scattering events are described by angular-dependent scattering coefficients in random media which follow from the Born approximation. In order to validate the method, the simulated envelopes are compared to average envelopes obtained by full waveform modeling with a finite difference method in two-dimensional random media with Gaussian and exponential correlation functions. Envelope shapes agree remarkably well for both short and long lapse times and for a broad range of scattering parameters. We conclude that the use of Born scattering coefficients in RTT does not pose severe limits on its validity range. Even in the strong forward scattering regime, envelope broadening and peak amplitude delays can be successfully modeled if one includes the wandering effect as obtained from the parabolic wave equation and Markov approximation into RTT.