Benchmark models were developed to evaluate the cold-critical start-up measurements performed during the fresh core reload of the neutron radiography (NRAD) reactor with low-enriched uranium fuel. Experiments include criticality, control rod worth measurements, shutdown margin, and excess reactivity for four core loadings with 56, 60, 62, and 64 fuel elements. The worths of four graphite reflector block assemblies and an empty dry tube used for experiment irradiations were also measured and evaluated for the 60-fuel-element core configuration. Dominant uncertainties in the experimental keff come from uncertainties in the manganese content and impurities in the stainless steel fuel cladding as well as the 236U and erbium poison content in the fuel matrix. Calculations with MCNP5 (Monte Carlo N-Particle version 5-1.60) and ENDF/B-VII.0 neutron nuclear data are ∼1.4% (9σ) greater than the benchmark model eigenvalues, which is commonly seen in Monte Carlo simulations of other TRIGA (Training, Research, Isotopes, General Atomics) reactors. Simulations of the worth measurements are within the 2σ uncertainty for most of the benchmark experiment worth values. The complete benchmark evaluation details are available in the 2014 edition of the International Handbook of Evaluated Reactor Physics Benchmark Experiments.