From a conditional expectation point of view, this chapter first introduces the definition of regressions where regression functions remain unknown for researchers, and then the classification of regressions is discussed. Importantly, the implementational procedure of series methods is elucidated through nonparametric regressions while that is similar in other situations. Two limit theorems about the series estimation for univariate nonparametric regressions with bounded and unbounded supports are established respectively that are benchmarks in comparison with other estimation methods in more complicated circumstances. Furthermore, some practical problems using series methods are also investigated, such as how to choose truncation parameter, and orthogonal series versus power series.