Despite enormous interest in recent years in the subject of the nurse-patient relationship, relatively little attention has been given in the nursing literature to how nurses manage their personal involvement in these relationships.This paper describes how the methods of grounded theory were used to explore the complex phenomenon of involvement.Through the methods of constant comparative analysis, theoretical sampling, mapping and theoretical memos, a clear distinction between involvement and over-involvement was discovered, and theoretical saturation was reached. Involvement emerged as therapeutic and beneficial, while over-involvement was revealed as inherently dysfunctional. In addition it was discovered that the two phenomena could be separated by their consequences, with the former resulting in positive outcomes for both nurses and patients, and the latter bringing negative repercussions for all concerned.The study resulted in the development of a theory of managing involvement, which concerns how nurses manage their personal involvement in relationships with cancer patients.