The paper applies a structural perspective to the analysis of political preferences. Examining two British surveys, the 1987 cross-section of the electorate and a panel survey that covers the 1983 and 1987 elections, the research explores the bases of persistent voting for the same party, location on left-right scales, and the probability of holding the same policy views on a host of different issues over time. A set of structural variables rests at the heart of the paper's theory: discussion networks, patterns of interactions with members of political parties, social class networks, and location in the social structure. Several hypotheses guide the analysis: The effects of the structural variables on the probability of casting a ballot for the same political party in any one election and in adjacent elections will remain, even after controlling for party identification; political party socialization; location on left-right scales; positions taken on any and all political issues; age, and past levels of electoral stability. The effects of structural variables on left-right position will remain, even after controlling for locations on alternative left-right scales. Finally, reinforcing attitudinal context provides the only consistent determinant of stable policy positions, after controlling for a host of alternative explanations including level of education; age; interest in politics, and a general propensity to offer stable answers to political questions.