Soil‐heat fluxes calculated using DeVries' theory (1958, 1963) were compared with those experimentally determined in a field of Avondale loam at Phoenix, Ariz., on 6 days representing different seasons of the year. A fair agreement between measured and computed fluxes was obtained only after modifying the air shape factor curve and ignoring heat transfer due to water vapor movement. The omission of the latter implied that heat transfer by pure conduction was most important and that thermal and isothermal vapor fluxes exactly cancelled during the day and were insignificant at night.“Measured” thermal conductivities were also determined from the ratio of the measured heat fluxes to the corresponding temperature gradient for those times when it was unlikely that isothermal vapor movement was significant. The lack of a temperature dependence in these data, as well as the flux comparisons, strongly indicate that the theory overestimated thermal vapor movement. These data plus others in the literature indicate that an individual “calibration” of the theory for a particular soil is required before reliable predications of soil‐heat flux can be obtained.