A stochastic model for delay-Doppler map (DDM) simulation from global navigation satellite system reflectometry (GNSS-R) systems is presented. The aim is to provide a useful tool for investigating the performance of estimation and retrieval algorithms that are based on finite time series. The scattering inside a delay-Doppler cell is modeled as the sum of a random number of contributions from inner specular points that, as the mean number of such contributions gets larger, tends to a compound-Gaussian process. The statistical averages reveal that the model is fully consistent with the previous results provided by Zavorotny and Voronovich. Numerical simulations of large airborne and spaceborne DDMs are easily practicable and show the clear patterns due to signal fluctuations and thermal noise that fade away when the number of averaged observations increases. Comparisons with TechDemoSat-1 data show that the model and the simulation scheme provide accurate realizations of the onboard-processed DDMs.