Objectives: This study had two aims. The first was to delineate the boundary between effective and inspiring teaching practices with cluster analysis again with a widely-researched instrument, International Comparative Analysis of Learning and Teaching (ICALT) (van de Grift, 2007; 2014) after initial success with Comparative Analysis of Effective and Inspiring Teaching (CETIT), a newly-developed classroom observation instrument that combined features of both effective and inspiring teaching (First Author et al., 2019) (Figure 1). Then, models hypothesizing some theoretical relationships of inspiring teaching behaviors were tested with structural equation modelling (SEM).Theoretical framework: Behavioral characteristics of inspiring teaching were initially characterized with a small sample by Sammons and colleagues (2014; 2016), but were found somewhat overlapped with those characteristics often associated with effective teaching.Methods: The hierarchical cluster analysis were conducted with SPSS 25 with five CETIT factors and seven ICALT factors. For SEM performed in LISREL 8.8, only the CETIT factors were used to explore the second aim. An inter-rater reliability over 0.7 (Gwet, 2014) was established using two different training videos before viewing the lesson samples.Data sources: The lesson sample was based on 260 videotaped lessons from both primary and secondary schools in Hong Kong, Shenzhen, and Guangzhou, selected from a pool of 500+ lessons in an earlier study by First Author and colleagues (2015). Table 1 summarizes the descriptive and correlation results.Results and conclusions: The hierarchical cluster analysis results showed two clusters identified (Figure 2). Average linkage to estimate the distance among factors was employed to overcome the shortcoming of single and complete linkage (Yim & Ramdeen, 2015). Together with the Agglomeration Coefficients, the results suggested location should stay at stage 10 with an elbow effect (Figure 3), where the factors