How well do members of the House adhere to the preferences of their constituents within standing committees? Roll call voting behavior on the floor largely reflects district preferences, but voting in committees is not easily monitored by constituents. This may allow legislators to disregard constituent preferences without suffering electoral consequences. I use a dataset of individual committee votes to create Optimal Classification scores for legislators within committee-congresses. Two-way fixed effects estimates and a redistricting natural experiment show that members of the House are largely responsive to district preferences, but the quality of representation declines over time as the district changes. Representation quality is also conditioned by committee heterogeneity and electoral security. In committees where jurisdictional complexity is high and party preferences are outlying, legislators’ voting records diverge from the preferences of their district, though low incumbent vote share strengthens the relationship between district preferences and committee voting.