The wireless medium has a time-varying feature due to fading, and individual users experience different degrees of fading. Opportunistic scheduling exploits such diversity of fading across users to improve network performance. In this paper, we explicitly analyze the capacity of opportunistic scheduling in various fading environments. To that end, we use Nakagami-m fading as a fading channel model and analyze the multiuser diversity capacity when considering the number of users and fading parameters. Our analysis shows that the mean capacity is explicitly decomposed into three factors: the average signal-to-noise ratio (SNR), the number of users, and the capacity improvement that comes from diversity of fading across users. Our analytical results provide the precise impact of fading channels on multiuser capacity, which asymptotic analysis cannot capture. We verify the analytical results via simulations.