The concept of deep geological disposal was developed in order to permanently remove radioactive wastes from the human environment. Retrieval of wastes was reckoned to be a scenario of such low probability that little effort was initially devoted to its study. In recent years, however, there has been an increasingly active debate on whether fully passive (and safe) systems provide a sufficient level of practicability of retrievability. This debate is linked directly to practical, technical matters, such as the design of the facility, the operating procedures and also the institutional programs throughout the lifetime of a repository. But there are also philosophical issues involved in addition to these purely technical issues.

For retrievability, the questions are: How easy does retrieval have to be in the different stages of repository development? What is the rationale for requiring retrievability? What technical measures and methods are feasible? Should specific features facilitating retrievability be introduced into the repository design? How do such measures impact on other aspects of system performance and on other issues (such as safeguards)? This paper addresses the questions raised and presents a subjective set of conclusions, as a means to stimulate further debate on this topical issue.