Primaries may contribute to polarization in other ways than a selective effect emanating from voters. This chapter tests a second potential mechanism of polarization, where incumbent members of Congress may respond to being challenged in a primary by adopting more liberal or conservative voting patterns in subsequent congresses. To test incumbents’ responses, it uses a series of fixed-effects models, clustered at the representative level, with roll-call movement as the key dependent variable. When considering the universe of all primary challenges, incumbents do not respond positionally, but when primaries are ideological and factional, they move toward their ideological pole. These effects are larger for factional primaries, indicating that incumbents are most responsive when a primary opponent has the support of an alternative party faction. These effects are larger for Republican than Democratic members of Congress, which is one way in which primaries may contribute to asymmetric polarization. These findings indicate that primaries may matter for polarization because incumbents believe them to be important and so are responsive to them.