Soil temperatures are closely coupled with soil moisture in permafrost environments. However, dynamic changes in soil moisture have not been given much attention in permafrost modeling, e.g. in long-term predictions. We illustrate the importance of surface soil moisture for permafrost modeling using the Cryogrid-3 model (a permafrost model with surface energy balance boundary condition) coupled with a dynamic soil moisture module. We explore the two key processes by which surface soil moisture affects soil temperature profiles: evapotranspiration and changing thermal properties. These two have partially opposing tendencies; evaporation cools moist soils; the larger thermal conductivity leads to increased heat flux to deeper layers and thus warmer deeper soil temperatures. The magnitude of the effect depends on the depth at which the temperature is measured, the environmental conditions and soil type in complex ways that are difficult to capture with traditional but widely used schemes such as those based on n-factors. We then show how permafrost modeling can benefit from soil moisture information by assimilating satellite soil moisture observations from Radarsat-2 into the Cryogrid-3 model. The assimilation exploits the dynamic coupling between surface soil moisture and the soil temperature profile to update the soil temperatures given a (noisy) soil moisture observation. The soil temperature estimates improve most for thick organic soils in our subarctic tundra site (Trail Valley Creek, NWT, Canada), but improvements are also found for mineral soil hummocks with thinner organic soil covers. The results highlight the importance of accurate soil moisture information for understanding and predicting soil temperatures in permafrost regions.