The article aims to explore the attributes of quality and quality improvement including the process and specific actions associated with these attributes--that contribute enhancing quality in Iranian Virtual Higher Education (VHE) institutions. A total of 16 interviews were conducted with experts and key actors in Iranian virtual higher education. A constant comparative analysis was adopted to construct a grounded theory model. Drawing on the experiences and perspectives of key actors and experts closely associated with quality in e-learning, a paradigm model for quality improvement in virtual higher education institutions was developed. The model articulates causal conditions, action/interaction strategies, consequences, contextual factors and intervening environments. Interestingly, quality of learning, i.e. deep learning was the core phenomenon in quality of virtual higher education institutions.