Typical cyber security risk assessment methods focus on the system under consideration, its vulnerabilities, and the resulting impact in the event of a system compromise. Cyber security, however, increasingly requires anticipating the moves of intelligent adversaries, who make decisions based on a range of factors including the cost of their attacks. A study of current risk assessment literature and industry practice shows that consideration of this cost is a notable gap in the understanding of adversaries. The factors of cost experienced by an adversary are established in this paper as Time, Finance, and Risk, supported by a practical study undertaken with relevant security practitioners. Using these factors as a base, a framework is proposed and developed to support the probabilistic determination of cost incurred by an adversary. This framework is an important extension to existing cyber security risk assessments, and is demonstrated in the paper through the use of a case study.