Discussion of the thermal conductivity of particulate materials is dispersed over several decades and a wide range of disciplines. In addition, there is some disparity among the reported values. This paper presents a review of the methodology available for the study of thermal conductivity of particulate materials, with an emphasis on low atmospheric pressures, and an assessment of the dependability of the data previously reported. Both steady state and nonsteady state methods of thermal conductivity measurement are reviewed, delineating the advantages, disadvantages, and sources of error for each. Nonsteady state methods generally are simpler and more efficient. The transient hot wire and differentiated line-heat source are the preferred methods for the laboratory. These methods are better suited for small samples and short measurement times and are therefore the best methods to use for a series of comprehensive studies. Results of previous studies are presented, compared, and evaluated. A good way to assess the relative accuracy is to compare the values of thermal conductivity versus atmospheric pressure obtained from several experimenters. The lowest values of thermal conductivity at vacuum and very low atmospheric pressure, and the steepest slopes on the thermal conductivity versus atmospheric pressure curves, are indicative of the most accurate data. Previous thermal conductivity studies have shown that the thermal conductivity of particulate materials increases with increasing atmospheric pressure, with increasing particle size, and with increasing bulk density of the material. At vacuum, the thermal conductivity of particulate materials is proportional to the cube of the temperature. The temperature dependence of thermal conductivity is much less obvious at higher atmospheric pressures.¿ 1997 American Geophysical Union