Behavioral self-regulation supports young children's learning and is a strong predictor of later academic achievement. The capacity to manage one's attention and control one's behavior is commonly measured via direct assessments of executive function (EF). However, to understand how EF skills contribute to academic achievement, it is helpful to investigate how EF manifests in the classroom context and in children's overt behavior. The current study observed 172 kindergarteners for a single school day and captured the total proportion of class time children were off-task in the classroom. This behavior was further classified into specific subtypes to assess whether these categorizations differentially predicted components of EF and academic achievement in first grade. Results indicated that children with lower response inhibition spent statistically significantly more time in one type of off-task behavior (i.e., off-task actively engaging with other materials), and children with lower working memory spent significantly more time in another type of off-task behavior (i.e., off-task passively disengaged). Higher proportion of class time spent off-task passively disengaged in kindergarten further statistically significantly predicted fewer gains in reading comprehension in first grade. These findings illustrate the utility of measuring children's EF in a classroom context, and how fine-grained observation systems can shed light on the specific classroom and child processes that influence learning.