North American Land Data Assimilation System (NLDAS) phase 2 has generated 31-year (1 January 1979 to 31 December 2009) water fluxes, energy fluxes and state variables from four state-of-the-art land models (NCEP/Noah, NASA/Mosaic, OHD/SAC, Princeton and Washington/VIC). However, until now, their soil temperature output has not yet been comprehensively evaluated. This study uses observed multiple-soil-layers monthly mean soil temperature over the 137 stations in the continental United States [Hu and Song, 2003] to evaluate Noah soil temperature simulation for the period from January 1979 to December 2002 because only Noah outputs the available soil temperature for phase 2. This evaluation includes assessment of simulation skill and analysis of error statistics for monthly and annual time scales. In addition, we used 3-year (1 January 1997 31 December 2000) hourly soil temperature observed from 72 sites over the Oklahoma Mesonet network to assess hourly and daily simulation skills and errors for 3 model soil layers in Noah (0-10 cm, 10-40 cm, and 40 -100 cm). Finally we show several examples comparing observed and simulated soil temperature associated with special weather events. This study objectively evaluates the Noah model ability to simulate soil temperature on multiple-timescales by comparing simulations with in-situ observations, and also indicates the shortcoming in both the Noah model and in the NLDAS forcing data sets.