In many species that fight over resources, individuals use specialized structures to gain a mechanical advantage over their rivals during contests (i.e. weapons). Although weapons are widespread across animals, how they affect the probability of winning contests is still debated. According to theory, understanding weapon function during contests is essential to: (i) understanding its importance in determining the winner, and (ii) identifying what weapon traits (e.g. weapon length versus shape versus performance) are most relevant for contest success. However, quantitative evaluations of how weapon function affects the extent to which weapon traits influence contest success are still lacking. Here, we first develop an individual‐based model to evaluate how increasing the influence of the weapon in determining the winner translates to differences between winners and losers. Then, we use a meta‐analysis to identify: (i) whether different weapon measures influence contest outcome differently; (ii) how animals use their weapons during fights – i.e. weapon function; and (iii) if weapon function correlates to how weapons influence contest outcome. Our model showed that, as weapons increased the chance of determining the winner, the mean difference between winners and losers also increased. Therefore, in our meta‐analysis we used the mean trait difference between winners and losers as a proxy for the extent to which weapons influence contest success. The literature search identified 49 suitable studies, containing information for 52 species, totalling 107 effect sizes. Four main patterns emerged. First, most of the literature focuses on linear measures of weapons, while performance measures are concentrated on Crustacea and Squamata; other types of measures were rare. Second, differences between winners and losers in linear measurements were greater than differences in performance measurements when all species were combined (and when we used only a subset). Third, species that bear weapons almost always perform visual/tactile displays before engaging in physical contact. And fourth, while the way individuals display their weapons did not influence the importance of weapon size on contest outcomes, fighting style predicted when differences between winners and losers would be higher. Species that used their weapons to push or lift (even when combined with other functions) showed greater differences between winners and losers when compared to species that used their weapons to impact, pierce, pull or squeeze. Overall, our results show that we have an incomplete understanding of animal weapons built mostly on weapon size and a few select taxa. Thus, we should start focusing on measuring weapons according to how they are used during contests and in a wider diversity of species. One way forward is to conduct studies that integrate weapon morphology to weapon function to ensure we are measuring the most ecologically relevant variables.