This study is framed by Bandura's social cognitive theory and Dewey's idea of habits and investigates types of science teachers based on four constructs: (a) self-efficacy beliefs, (b) twenty-first-century learning attitudes, (c) the frequency of implementing inquiry-based instruction, and (d) the frequency of using technology in teaching. K-means clustering analysis was run on a sample of 837 US elementary and secondary science teachers. Multinomial logistic regression was then used to characterize each group using variables such as years of teaching experience, National Board Certification (NBC) status, gender, grade-level taught, school location, STEM career awareness, and leadership attitudes. We found four science teacher groups and named them the idealist, moderate, reform-minded, and generalist. Further analyses revealed that gender and geographic location did not significantly influence cluster membership. However, we found that clustering was significantly associated with NBC status, grade-level taught, years of teaching experience, STEM career awareness, and leadership attitudes. Notably, the reform-minded science teachers were more likely to hold an NBC status, teach secondary science, and have a high awareness of current STEM careers. Moreover, the generalist science teachers were those teaching elementary levels and without NBC status. These findings inform the design of tailored professional development programs targeting specific science-teacher characteristics.