We demonstrate that high-quality total radiation width ( γ ) data are a virtually untapped resource for testing and improving nuclear models. To this end, we obtained unprecedentedly large sets of γ values for all six sand p-wave J π values for 95 Mo neutron resonances. We show that γ distributions simulated in the framework of the nuclear statistical model are in sharp disagreement with the data. Simulations modified to include doorway effects resulted in much better agreement. These results call into question the reliability of the nuclear statistical model. There has been significant controversy surrounding γ decay of molybdenum isotopes below the neutron separation energy. Widely different shapes for the photon strength function (PSF), representing the decay strength as a function of γ -ray energy, have resulted from various measurements [1][2][3]. In addition, the enhancement in strength at low γ -ray energies reported for molybdenum isotopes [1,2] has been called into question [4,5], although more recent data [6] strongly support its existence. There is also disagreement about the nature of γ decay following neutron capture on molybdenum isotopes. On the basis of a very few measured total radiation widths γ , it was suggested [7] that significant nonstatistical effects affected this decay. In contrast, it was found that the extreme statistical model was in very good agreement with two-step cascade γ spectra following thermal neutron capture [4] as well as coincidence-pulse-height spectra and other data for neutron-resonance capture [5].Resolving these controversies is of central importance for applications such as advanced nuclear fuel cycles [8] and astrophysical nucleosynthesis studies. For example, the PSF shape can have substantial impact on astrophysical reaction rates [9] in explosive environments, and therefore significantly impact the resulting nucleosynthesis. In addition, as the nuclear statistical model is used to calculate almost all rates for nuclides beyond the reach of current experiments, it is important to test basic assumptions of this model.Total radiation widths of neutron resonances represent a virtually untapped source of data in this endeavor. Due to the very large number of levels below the neutron binding energy B n , there are thousands of channels by which γ decay can occur following resonance neutron capture, a few of which are schematically illustrated in Fig. 1. Hence, energies E γ of these transitions span the entire range of level density (LD) and PSF below B n . The total radiation width γ is the sum of all the partial widths for the initial (primary) γ transitions from the resonance.According to the statistical model, strengths of each of these primary transitions follow the Porter-Thomas distribution (PTD) [10], with average strength proportional to the PSF * Corresponding author.(with an appropriate energy factor, which is E 3 γ for dipole transitions) at the corresponding energy. It is further assumed that all, or at least the vast majority, of the channels are indepen...