Japan Atomic Energy Agency (JAEA) and Los Alamos National Laboratory (LANL) jointly developed the Epithermal Neutron Multiplicity Counter (ENMC). A measurement test was performed using the standard samples and its results showed that ENMC achieves high measurement accuracy (approx. 0.4\%) for $^{240}$Pu effective mass under the optimum conditions. However, in the practical measurement for nuclear material accountancy or safeguards, a bias is observed due to the variation of the sample properties. With this recognition, JAEA jointly with LANL conducted simulations for identifying the causes of this bias. The simulation results showed that the dominant cause of the bias is variation in sample density and this bias can be mitigated by correcting neutron counting efficiency. JAEA and LANL evaluated the applicability of correction methods for the neutron counting efficiency by real measurement data and by simulation data. For the real measurement, the results showed that the real measurement data is difficult to be applied to the correction because of its significant measurement error. For the simulation, we evaluated the neutron counting efficiencies for typical density of MOX pellet and powder. Consequently, total measurement uncertainty for Pu mass quantification by using combination of ENMC and NDA for isotopic ratio of Pu (HRGS) attains 0.7\% which is equivalent to the destructive assay level.