According to the stereotype content model (SCM), warmth and competence are the two universal dimensions of social cognition. While the relationship between warmth and competence is confused. Some studies suggested that warmth judgments carried more weight in social cognition and there was an effect of the primacy of warmth, but some were not. Moreover, it is unknown that what is the relationship between warmth and competence in facial attention processing. Therefore, the aim of the present study is to explore the relationship between warmth and competence in face processing from the perspective of attention. Thirty participants (seventeen females) participated in the experiment, which was designed a 2 (facial type: warmth, competence) × 2 (valance: positive, negative) within-subject study. And this experiment recorded the time to first fixation, the fixation counts and the total fixation duration by eye tracking technique. During the experiment, the two sides of the screen would present warm and competent faces randomly. And then a black probe point would appear randomly on the left or right side. The participants were told that they should judge the location of the probe as accurately as possible. There was a total of 270 trails, and all faces were presented only once, and the positions were balanced. At the end of the experiment, the participants were asked to evaluate the approach-avoidance of all faces with a 9-point rating (1 = very want to avoid, 9 = very want to approach). In terms of eye tracking data, the study found that the time to first fixation on the warm faces were shorter than the competent faces. Compared to the competent faces, fixation counts are more and total fixation duration are longer on warm faces, and similarly the fixation counts and total fixation duration of positive-judgment faces are more than the negative-judgment faces. The results of approach-avoidance judgment are significant between the evaluation of the negative warm and competent faces; individuals are more likely to avoid negatively warm faces than negatively competent faces. And the positive dimension was not significant. In the early stages of facial attention, the time to first fixation on the warm faces was shorter, which indicated that the judgment of others’ intentions takes precedence over the judgment of the competence to achieve intentions in the early stages of face perception, capturing the observer’s attention more quickly. In the late stage of attention, the attention to the warm faces are maintained. Compared with the competent faces, the warm faces may convey stronger threatening information. What is more, individuals evaded judgments about warm faces more than competent ones under negative conditions, which might reflect the intention of others directly affects the individual’s survival value and interpersonal activities, and then the individual present the alertness to the warm faces. Therefore, the data supported that there exists an effect of the primacy of warmth during the face processing.