Ambivalent attitude as a topic in the field of attitude research is newly rising up. Ambivalent attitude is defined as the simultaneous existence of positive and negative beliefs or emotions with regard to the same object in an individual’s attitudinal basis. Previous studies investigated whether ambivalent attitude would facilitate or inhibit the processing of information which was related to the attitude objects, but the results are controversy. To address this question, the present study examined such effects from both explicit and implicit ambivalent attitude levels. The experiment was carried out as following procedures: First, we surveyed the ambivalent attitude among college students by open-questionnaire. Descriptive statistics showed that “EXERCICE” is an attitude objects which can generally induce ambivalence. Second, in formal experiment, participants who were random chosen and did not take part in pre-test were asked to perform two task presented on computer: a split semantic differential measure which was used for testing explicit ambivalent attitude induced by the attitude objects and an Extrinsic Affective Simon Task (EAST) which aimed to explore the implicit ambivalent attitude induced by the attitude objects. The sequence of tasks was counterbalance between participants. And then the participants were divided into higher and lower explicit ambivalent group, higher and lower implicit ambivalent group according to their scores in the split semantic differential measure and EAST. Then, the Study Test paradigm was used to probe the effects of explicit and implicit ambivalent attitude on the processing of information related to “EXERCICE”. In this paradigm, the unintentional and deliberately processing ways were also manipulated. The results indicated that (1) the response latency of the higher explicit ambivalent group was saliently shorter than the lower explicit ambivalent group. (2) The response latency which the higher implicit ambivalent group used to process deliberately the attitude objects information was significantly longer than the lower implicit ambivalent group. But in the unintentional processing level, the two groups did not manifest any difference. (3) The response latency which the higher implicit ambivalent group used to process information unrelated to the ambivalent attitude objects was conspicuously longer than the lower implicit ambivalent group. However, the response latency used for processing information related to the ambivalent attitude objects did not differ between two groups. These results suggested that implicit and explicit ambivalent attitude could influence processing of information associated this attitude in distinct ways. And the intensity of ambivalent attitude and the ways of information processing could affect information processing related to the ambivalent attitude objects. These results were further discussed in the sense of dual-attitude model.