In designing a heterogeneous reactor it is not enough to be able to calculate the nominal temperature of the fuel elements; one must be able to calculate the probability that the surface temperature is either less than a given value or lies between given limits. This paper presents a general method of analyzing this problem and applies the method to the particular case of a heterogeneous, gascooled reactor. It is shown that one need not assume each statistical variable controlling the temperature to be normally distributed; the individual variables can have any distribution. For design purposes, however, one generally must assume that any value of the parameters, between fixed limits, is equally likely, and for this case it is shown that the fuel element surface temperature itself will be adequately approximated by a normal distribution even though the independent variable has a rectangular frequency function.