Monitoring and verification o f t he n uclear f uel c ycle r equires t echnology a t e very s tep including power generation. Monitoring the count rate from a nuclear reactor core in neutron detectors is an effective way to confirm the power level as well as startup and shutdown ac tivities. For low power facilities, the steady state count rate in sub-critical and delayed critical states can be indistinguishable. Temporal analysis of fission c hain r eactions, a lso k nown a s “noise” a nalysis, c an d istinguish the reactor kinetic state. The CROCUS zero-power reactor has been studied comprehensively with detectors measuring the time-correlated noise of the reactor core at critical. In this work, we place two trans-stilbene detectors set in the CROCUS moderator to measure the prompt neutron decay constant (α) at critical and in several sub-critical states. We conduct four, two-hour measurements at the critical water level (960 mm, p = 0 $), and three sub-critical water levels (800-900 mm corresponding to p = [−0.5, −0.9, −1.4] $ respectively; p estimates are derived from Serpent 2 simulations in prior work). We analyze the (γ,γ) cross power spectral density distribution from each measurement and calculate values of α = [151.8 ± 0.7, 213.6 ± 1.7, 264.5 ± 2.8, 333.0 ± 5.0] s−1 for the p = [0, −0.5, −0.9, −1.4] $ cases respectively. Additionally, we extrapolate the α value at critical from sub-critical α estimates with a linear fit dependent on p yielding α = [150.4 ± 2.1] s−1