This volume constitutes the thoroughly refereed post-conference proceedings of the Second International Workshop on Graphical Models for Security, GraMSec 2015, held in Verona, Italy, in July 2015. The 5 revised full papers presented together with one short tool paper and one invited lecture were carefully reviewed and selected from 13 submissions. The workshop contributes to the development of well-founded graphical security models, efficient algorithms for their analysis, as well as methodologies for their practical usage, thus providing an intuitive but systematic methodology to analyze security weaknesses of systems and to evaluate potential protection measures.