Yes. Normally, folks "couple" a thermal study to a static study, then the static study is fed into the fatigue study and cycled there. Assuming 'worst case' thermally-induced stresses occur at thermal equilbrium for the "hot" case, and another set of thermally-induced stresses occur at thermal equilibrium for the "cold" case (i.e., the zero-strain temperature is somewhere in between the two thermal extremes), you could do this:
Run the thermal analysis for heating up, then create a static study to read the temperature field. Next, create a thermal study for cooling down, then create a static study to read the thermal temperatures from the cooling down study. So you will have two static studies, one for each extreme. Then create a fatigue study and you can instruct it to use the results from both of the static studies and "cycle" them. Hope that helps a little.