The thermal conductivity of the thin seasonally freezing and thawing soil layer in permafrost landscapes exerts considerable control over the sensitivity of the permafrost to energy and mass exchanges at the surface. At the same time, the thermal conductivity is sensitive to the state of the soil, varying, for example, by up to two orders of magnitude with varying water contents. In situ measurement techniques perturb the soil thermally and are affected by changes in soil composition, for example through variations in thermal contact resistance between sensor and soil. The design of a sensor for measuring the temperature of the soil rather than the axial heating wire temperature has consequences for the modeling of heat flow. We introduce an approximation of heat flow from a heated cylinder with thermal contact resistance between the cylinder and the surrounding medium. This approximation is compared to the standard line source approximation, and both are applied to data measured over a one-year period in northern Alaska. Comparisons of thermal conductivity values determined numerically using the line source solution, line source approximation and the analytical form of the heated cylinder model fall within 10% of accepted values, except for measurements made in pure ice, for which all methods of calculation under-predicted the thermal conductivity. Field data collected from a complete freeze-thaw cycle in silty clay show a seasonally bimodal apparent thermal conductivity, with a sharp transition between frozen and thawed values during thaw, but a three-month transition period during freezing. The use of soil composition data to account for changes in heat flow due to the effect of latent heat during phase change results in a relationship between soil thermal conductivity and temperature. © 2006 Elsevier B.V. All rights reserved.