As females of many species mate with more than one male, ejaculates often face competition from the sperm of other males. In recent years, numerous papers have been published on theoretical predictions of evolutionary, behavioural and physiological responses to variation in the strength of sperm competition (SC). These theoretical predictions have also been extensively tested. However, although predictions from SC theory are relatively straightforward, extra caution has to be paid in the design of experiments testing them. One difficulty is for example to disentangle immediate and mean SC risk and intensity. Without carefully designed experiments, it is also very easy to simultaneously increase SC risk and the probability of intense SC - a situation for which we currently have no clear predictions, as the theoretical models to date only assume variation in either SC risk or intensity. In this paper, we discuss these and some other pitfalls related to manipulations of SC risk and intensity and suggest how to avoid them.