Generalized anxiety disorder (GAD) is common during the perinatal period. Since its inception, substantial efforts have been put forth to improve the specificity and interrater reliability of GAD. More recently, increased attention has been given to the behavioral features of GAD, with evidence to suggest that hypervigilance, checking, and avoidance may be particularly relevant. It is unknown however if these behaviors lead to improved classification and understanding of GAD. The present study examines: (1) the proportion of perinatal individuals with and without GAD who endorse hypervigilance, checking and avoidance; (2) interrater reliability of GAD during the perinatal period; (3) whether inclusion of hypervigilance, checking, and avoidance in GAD diagnostic criteria leads to improved interrater reliability; and (4) if hypervigilance, checking and avoidance significantly predict GAD diagnostic status beyond current features of the disorder. Thirty-eight perinatal women, who were predominantly white and highly educated were randomized to one of two assessors to complete a semi-structured diagnostic assessment. Interviewers queried about the presence of current mental health disorders and engagement in behaviors of interest. Each assessment was independently rated by three assessors. More participants with GAD reported engaging in hypervigilance, checking, and avoidance than those without GAD. Interrater agreement of GAD diagnoses was excellent (κ = 0.91). Inter-rater agreement of GAD diagnoses was comparable (κ = 0.92) when checking was included in the diagnostic criteria of GAD. Checking also significantly predicted GAD diagnostic status beyond existing features of the disorder. These results support continued evaluation of the role of checking in GAD.