We describe new dosimetry (radiochemical) ENDF evaluations for yttrium, iridium, and thulium. These LANL2006 evaluations were based upon measured data and on nuclear model cross section calculations. In the case of iridium and yttrium, new measurements using the GEANIE gamma-ray detector at LANSCE were used to infer ( n , x n ) cross sections, the measurements being augmented by nuclear model calculations using the GNASH code. The thulium isotope evaluations were based on GNASH calculations and older measurements. The evaluated cross section data are tested through comparisons of simulations with measurements of reaction rates in critical assemblies and in Bethe sphere (sometimes called Wyman sphere) integral experiments. Two types of Bethe sphere experiments were studied – a LiD experiment that had a significant component of 14 MeV neutrons, and a LiD-U experiment that additionally had varying amounts of fission neutrons depending upon the location. These simulations were performed with the MCNP code using continuous energy Monte Carlo, and because the neutron fluences can be modeled fairly accurately by MCNP at different locations in these assemblies, the comparisons provide a valuable validation test of the accuracy of the evaluated cross sections and their energy dependencies. The MCNP integral reaction rate validation testing for the three detectors yttrium, iridium, and thulium, in the LANL2006 database is summarized as follows: (1) ( n , 2 n ) near 14 MeV: In 14 MeV-dominated locations (the LiD Bethe spheres and the outer regions of the LiD-U Bethe spheres), the ( n , 2 n ) products are modeled very well for all three detectors, suggesting that the evaluated 89Y ( n , 2 n ) , 191Ir ( n , 2 n ) , and 169Tm ( n , 2 n ) cross sections are accurate to better than about 5% near 14 MeV; (2) ( n , 2 n ) near threshold: In locations that have a significant number of fission spectrum neutrons or downscattered neutrons from 14 MeV inelastic scattering (the central regions of the LiD-U spheres and the fast critical assemblies), the ( n , 2 n ) products are overpredicted by 5–30 % for the three detectors, suggesting either the threshold region ( n , 2 n ) cross sections are too high, or that the MCNP-simulated neutron flux is too large for neutron energies above about 8 MeV; (3) Capture: The capture products for yttrium are modeled accurately for the LiD Bethe spheres, but are underpredicted by about 20% for the LiD-U Bethe spheres and the critical assemblies; for iridium-191 they are predicted accurately in the critical assemblies; and for thulium they are generally overpredicted by 10–30 %; (4) Inelastic scattering in iridium: The evaluated 193Ir ( n , n ′ ) 193 m Ir cross section performs well over a very wide range of neutron spectra (where the 193m/190 spectrum hardness index varies by over three orders of magnitude), the differences between simulation and experiment typically being better than 10–15%; (5) Iridium 193m/190 spectrum hardness index: Our simulations reproduce the measured 193m/190 data typically to better than 10–20% over three orders of magnitude in the 193m/190 ratio. The aforementioned indications from data testing involving assemblies containing actinides - that the ( n , 2 n ) products are overpredicted by 5–30% – could be used to motivate a decrease in the evaluated ( n , 2 n ) cross sections in the approximately 8–12 MeV range. However, at this stage we have not modified these cross sections since: (a) They are consistent with the cross section laboratory measurements; and (b) It is possible that the cross sections are correct and instead the simulated integral assembly neutron spectrum is too high for neutron energies above 8 MeV. The latter possibility is particularly intriguing given all three detector materials showed a bias in the same direction, and that the evaluated actinide prompt fission spectra and inelastic scattering data are probably uncertain to at least 20% above 8 MeV. We also discuss refinements needed in the transport methods to faithfully represent the evaluated data.