Physical aggression (PA) and relational aggression (RA) are two major types of aggression. Both forms of aggression are associated with multiple psychosocial maladaptation for both attackers and victims, and has long been concerned by theoretical and empirical researchers. Researches demonstrated medium to high degree of correlation between PA and RA, indicating that PA and RA are co-occurring. The extant studies mainly employed latent class growth modelling to examine the associations between developmental trajectories of PA and RA, and provided insights for the co-occurring nature of PA and RA. However, these studies failed to reveal co-occurring patterns of PA and RA, which is better to be identified based on their co-occurring levels. Further, research indicated differential developmental trends of PA and RA, and their correlations varied across age, which suggested that the co-occurring patterns of PA and RA may demonstrate variation across ages. The broad aim of the present study was to identify the heterogenous co-occurring patterns of PA and RA using latent profile analysis (LPA), and their stability across ages. We would also examine the associations between co-occurring patterns with peer acceptance and rejection, by which to provide validity evidence for the co-occurring patterns. A total of 1967 seven-grade (mean age 13.22 ± 0.36 years old, 1018 male) students were investigated and followed for three years. Children’s teachers reported their aggression on eight items of PA and six items of RA from a Chinese version of Child’s Social Behavior Questionnaire. A standardized peer-nomination procedure was used to collect information on children’s peer acceptance and rejection. The measures showed good psychometric properties. The following results were found. (1) Five heterogeneous co-occurring PA & RA patterns were consistently identified at three time points, namely high PA & RA group, moderate PA& RA group, low-to-moderate PA and low RA group, low PA & low-to-moderate RA group, and low PA & RA group. (2) The memberships of adolescents belonging to each co-occurring PA & RA group demonstrated differential stability across ages. Adolescents’ memberships of low PA& RA group were most stable, and followed by high PA & RA group, and adolescents in low PA and low-to-moderate RA group showed the highest transition probability across age. (3) High PA & RA group had the highest level of peer rejection, and the lowest level of peer acceptance, while low PA & RA group demonstrated the opposite in terms of peer rejection, and acceptance, which were consistent with the social information processing deficits model on aggression and peer relationships, and provided evidence for the validity of the heterogenous co-occurring PA & RA groups. In conclusion, the current study demonstrated consistent co-occurring PA & RA patterns during early adolescence which were presented as heterogenous groups of different levels of PA and RA. Further studies on the nature of the co-occurring patterns and co-occurring PA & RA patterns across longer developmental periods were needed.