Solid state photosensors, usually called SiPM or G-APD, seem ideal devices to be used in Imaging
Atmospheric Cherenkov Telescopes (IACT). Nevertheless, their temperature dependence poses
questions about their suitability in the harsh environment intrinsic to the operation of IACTs.
While detailed measurements in the laboratory are possible with some sample sensors, limited
data about the performance and uniformity of large samples exist. The First G-APD Cherenkov
Telescope (FACT) is pioneering the usage of SiPMs for IACTs. Its camera consists of 1440
SiPMs and it is operated since October 2011 each night when observation conditions permit.
Using no temperature stabilization system for the sensors, their temperature is closely coupled
to the outside temperature that can change by more than 20 $^\circ$C. While the strong temperature
dependence of the gain of the sensors was shown to be easily compensated by adapting the applied
voltage, there could also be higher order temperature dependencies of parameters like optical
cross-talk, after-pulsing and wavelength dependent photon-detection efficiency. While external
calibration devices could be used, one would have to proof that these devices do not have their own
temperature dependencies. Instead, we use the constant flux of high energetic cosmic ray particles
as calibration device. Their measured flux can depend on variable absorption and scattering of
Cherenkov light e.g. due to dust and clouds, as well as on seasonal variations of the atmosphere.
Nevertheless, using data sets where the temperature drastically changed within short time periods,
we show that temperature dependencies of FACT, including the SiPMs, are well under control.