A new calculation code, the Hot Spot Probabilistic Evaluation Code (HOSPEC), is presented for evaluating hot-spot factors in a fast reactor. This code calculates the probability distribution of temperature in the whole core by means of a Monte Carlo method. Each Monte Carlo trial involves a complete thermal conduction analysis, thereby reducing errors due to assumptions in analytic procedures currently in use. With this code it is possible to determine the probability that fuel pellets, fuel pins, or subassemblies will exceed the limiting temperature, as well as determine the number of such “hot spots” that will develop. A quantitative comparison is made of the results obtained from this code with those from other analyses of a prototype fast reactor. The comparison has indicated, among other points, the following findings: 1. For zero hot spots, a conventional analytic evaluation code SHOSPA gives conservative results, i.e., ∼20°C at the fuel center, at a 3σ confidence level. 2. It is of crucial importance to take into account the temperature dependence of the properties of the materials. Neglecting such dependence leads to a much more conservative temperature prediction, e.g., ∼50°C at the fuel center.