This study was conducted to test the following two hypotheses: (a) Interpersonal attraction was a linear function of opinion similarity, and (b) a recency effect would exist in a sequential presentation of similar and dissimilar opinion stimuli. In a 5 × 2 factorial design, 5 levels of opinion similarity (.10, .30, .50, .70, and .90,) and 2 opinion sequences (similarity-dissimilarity and dissimilarity-similarity) were employed. Each of the 100 subjects participating in this study interacted with a confederate in one of the ten experimental conditions. Significant effect on attraction responses was found only for the opinion similarity variable with no interaction between opinion variable and presentation sequence variable noted. Three explanations were suggested for the absence of order effect: (1) The opinion items or the experimental context had not made the subjects feel that their opinions on the items were especially important; (2) agreements and disagreements between the subject and the confederate were not on the same issue; and (3) subject's evaluations of inferred attributes lead to the formation of favorable/unfavorable affective responses, whereas similarity per se was not the real important determinant of those responses. The implication of these findings to similarity-attraction theories was also discussed.