Examined context (CX) effects on answers to attitude questions by surveying 1,174 respondents about 6 target issues. Before the target question, Ss were asked about either 2 related or 2 unrelated (neutral) CX issues. CX effects were larger when Ss reported that their beliefs about the target issue were both mixed and important. Coding of open-ended responses to follow-up questions suggested that the CX items may have changed how Ss thought about the target issues. Results support the accessibility hypothesis, which predicts that CX effects will be largest when respondents have well-developed, conflicted views about the target issue and when CX and target items are near each other in the questionnaire. (PsycINFO Database Record (c) 2007 APA, all rights reserved)