Abstract: We show that comparing geometrical algorithms, in general, is very hard to do in a fair way. Our focus here is on the existing ray-triangle intersection test routines. A set of rules are developed so that a fair comparison can be produced, and this is all implemented in an evaluation framework. If all algorithms are evaluated using different hit rates, on different machines, for different input and output data, and different compilers, then it is impossible to single out one "best" algorithm. However, we provide ranking tables of the algorithms as well as top-five lists for all test computers, which can help a reader pick a good algorithm. Furthermore, a program called a "ray-triangle advisor" is developed that people can use in order to determine which algorithm works best under their particular circumstances. We hope that our work can be useful for both developers of new intersection algorithms as well as for users of ray-triangle intersection algorithms.