We investigate the parameterized complexity of strategic behaviors in generalized scoring rules. In particular, we prove that the manipulation, control (all the 22 standard types), and bribery problems are fixed-parameter tractable for many generalized scoring rules, with respect to the number of candidates. Our results imply that all these strategic problems are fixed-parameter tractable for many common voting rules, such as Plurality, r-Approval, Borda, Copeland, Maximin, Bucklin, Ranked pairs, Schulze, etc., with respect to the number of candidates.