This paper describes the validation and sensitivity analysis of a process-based simulation model (BACSIM) for the control of beet armyworm, Spodoptera exigua, with baculoviruses. Model predictions are compared to results of independent greenhouse experiments in which second, third, or fourth instar larvae of S. exigua in chrysanthemum plots are treated with different concentrations of Autographa californica multicapsid nucleopolyhedrovirus (AcMNPV) and S. exigua MNPV (SeMNPV), two viruses with distinct differences in infectivity and mean time to kill. BACSIM provides robust predictions for the control of S. exigua populations in greenhouse chrysanthemum with both AcMNPV and SeMNPV. Mortality levels caused by AcMNPV and SeMNPV were generally predicted within a 25 percent margin of error compared to the observed values. None of the deviations was higher than 40 percent. All values of simulated foliage consumption, caused by S. exigua populations treated with AcMNPV or SeMNPV applications, fell within 95 percent confidence intervals of measurements. Simulated time to kill was, in general, lower than the measurements. This discrepancy may be caused by the behavior of S. exigua larvae which feed on the underside of chrysanthemum leaves where they are protected from polyhedra. This suggests that the larval foraging behavior may play an important role in the efficacy of baculovirus applications and should be further studied experimentally. This validated model can be used for the pretrial evaluation of the efficacy of genetically modified baculoviruses as biological control agents and for the optimization of spraying regimes in chrysanthemum cultivation.