The theory of a method of determination of reactivity from power spectral density measurements with 252Cf and the results of experiments with a critical assembly mockup of a liquid-metal fast breeder reactor (LMFBR) and with uranium (93.2 wt% 235U) metal cylinders and a sphere are presented. This method of reactivity determination has an advantage over existing methods in that it determines the reactivity only from properties of the reactor at the subcritical state of interest and thus does not require a calibration near delayed criticality. In these experiments, the reactivity was varied by changing the fissile loading or the amount of neutron absorber inserted; for the LMFBR mockup, the reactivity varied to ∼75 dollars subcritical, and for the uranium metal assemblies to ∼30 dollars subcritical. These experiments verified for the first time the predictions of theory that could be tested in the measurements. This method has potential use in the fuel loading of reactors to determine the reactivity far subcritical before initial criticality is achieved. It has the advantage of not requiring a calibration at known reactivity by another method; furthermore, the interpretation of the measured data to obtain the reactivity does not depend on relative or absolute values of the source intensity or detection efficiency. It can also be used to determine the reactivity of assemblies where loading to criticality is undesirable or where sufficient material to achieve criticality is not available.