The literature on separation anxiety disorder (SAD) presented two contentious issues relating to its assessment. First, studies are scarce in assessing the symptom structure of DSM-5 SAD among the adult population. Second, the accuracy in assessing the severity of SAD through measuring the intensity of disturbance and the frequency of occurrence of symptoms is yet to be studied. To address these limitations, the present study aimed to: (1) examine the latent factor structure of the newly developed separation anxiety disorder symptom severity inventory (SADSSI); (2) evaluate the necessity of using frequency or intensity formats through comparison of differences in the latent level; and (3) investigate SAD latent class analysis. Utilizing 425 left-behind emerging adults (LBA), the findings showed that a general factor with two dimensions (i.e., response formats) measuring frequency and intensity symptom severity separately has excellent fit and good reliability. Finally, the latent class analysis yielded a three-class solution best fitting to the data. Overall, the data provided evidence for the psychometric soundness of SADSSI as an assessment tool for separation anxiety symptoms among LBA.