Historically, attitudes have been hypothesized to bias memory in  a congenial manner.  Pro-abortion attitudes should thus be related to  greater memory for new (or experimentally provided) pro-abortion  statements than for anti-abortion statements (and vice versa for  anti-abortion attitudes).  This colloquium presents the results of a  series of primary experiments designed to test this hypothesis as well as  its reverse that attitudes sometimes bias memory in a non-congenial  manner.  Meta-analytic analysis of the attitude-memory literature  revealed a number of moderators of the congeniality effect; most notably,  accuracy-motivated and defense-motivated processing.  A series of primary  experiments validated aspects of the hypothesis that attitude-memory  effects (whether congenial or non-congenial) are  indeed moderated by  these two forms of motivational goal.  In addition, the studies show the  importance of taking internal attitude structure and/or attitude basis  (affective v. cognitive) into account;  when attitudes are primarily  affect based, memory is greater for congenial information.  When  attitudes are primarily cognitively based, however, the reverse  hypothesis of greater memory for non-congeniality is confirmed.  Two  remaining studies show that (1) instructing participants to focus on  their affective (vs. cognitive) responses to attitude-relevant statements  produces the same results as when individual difference measures of  attitude-basis are measured instead; and (2) the magnitude of the  attitude-memory effect can be decreased to minimal levels when  participants must process attitude-relevant information under cognitive  load.  Ongoing studies (without results) are designed to explore more  fully the mediational processes that produce the congeniality and  non-congeniality effects we have observed in this research.