This paper argues that Grice’s theory can still explain metaphors if we make the following revisions. The first revision to make is that not only flouting maxims but also observing maxims should explain metaphors depending on different types of metaphors. Secondly, we should admit the fact that sufficiently rich contexts can obviate the need to calculate maxims to arrive at metaphorical meanings. These revisions can address why sometimes no processing differences are found between literal languages and metaphorical languages even though ordinarily it is the case that metaphorical languages take a longer time to process than literal languages. Furthermore, the proposal of this paper can explain why category mismatches are not the necessary condition to evoke metaphors. Claiming that there are ways to make Gricean approach to metaphors still work, this paper also discusses high linguistic values of Grice’s maxims in comparison to other attempts to reduce the maxims, such as Horn’s two principles and the relevance principle. This paper concludes that we need to give more credit to Grice’s work not only for metaphors per se but also for diversity of the system.