In parts of the United States, in Soil Survey the mean annual soil temperature (MAST) is often estimated from the mean annual air temperature (MAAT) by adding 1°C. The objective of this study is to evaluate the relative difference between the MAST and the MAAT at various latitudes, soil types and moisture contents over an extended period of record. Soil climate monitoring stations were established throughout the United States (http://www.wcc.nrcs.usda.gov/scan/ ) and the Antarctic (http://soils.usda.gov/survey/scan/antarctica/index.html ). Hourly air and soil temperatures (at 50 cm depth) were collected at 30 stations for a period of record ranging up to 10 years. Seven stations were located in the McMurdo Dry valleys region of Antarctica. For the entire period of record at a station, the average MAST was subtracted from the average MAAT. The average difference was 1.72°C (±1.08) with a minimum value of -0.37 and maximum value of 3.84. The median value was 1.74°C. Soil properties range from 1 to 47% clay, 1 to 95% sand, and 0 to 33% organic C. The difference between MAST and MAAT was not significantly correlated to latitude, total clay, total sand, bulk density, soil moisture, or organic C contents. The highest correlation was with organic C (r = 0.28). The best general linear model using soil moisture, OC, total sand, and bulk density could only explain 20% of the variation in the difference between MAST and MAAT. Within the US, latitude was significantly correlated to the MAST (r = 0.73) as expected. The soil temperature sensor accuracy is ±0.6°C and the air temperature sensor accuracy ranges from ±0.2 to 0.4°C depending on the temperature. Sensor error can be up to ±1°C confounding the ability to explain variation in differences between MAST and MAAT. Sensor accuracy needs to be more controlled in future studies in an attempt to explain the variation between MAAT and MAST.