A method of monitoring stability of boiling water reactors (BWRs) has been developed. The stability parameters were derived from empirical discrete-time modeling of process noise signals and neutron noise signals. Data were taken from an operating BWR-4, and used to perform univariate analysis of average power range monitor (APRM), and local power range monitor signals, and multivariate analysis of APRM and the process signals, reactor pressure, and core flow rate. The parameters such as decay ratio, damping ratio, and characteristic frequency of oscillation, which represent the system stability, were estimated from the impulse response of the system. The impulse response was determined by using the time series models and contains information about the closed loop dynamics of a BWR. The results indicate the feasibility of using APRM noise analysis for monitoring overall core stability and temporal variations in the stability margin of the reactor. Any significant variation in the stability parameters can be studied using multivariate noise signal algorithms, and cause and effect relationships can be obtained. Because the derived parameters depend on the random noise properties of the signals, this nonperturbing method is most useful for monitoring changes in stability. If an absolute measurement is necessary, a perturbation test must be performed.