A reactor can be analyzed as a multiplicative stochastic process or, approximately, as a deterministic process. When feedback is present, the stochastic and deterministic analyses can differ qualitatively as well as quantitatively, as is illustrated by the concept of stability. In the present study, a stochastic model of a nuclear power reactor with 135Xe, 135I, and control feedback is considered as an example of a nonlinear stochastic process. The values of variances and covariances are calculated from the first- and second-moment equations, using an iterative procedure. Numerical criteria for the value of the feedback coefficient for marginal stationarity of the stochastic model are compared with the corresponding criteria for the stability of the corresponding linearized deterministic model and found to be identical, within eight significant figures.