The origins and types of secrecy are discussed in terms of an epigenetic model. The interplay of intrapsychic and social factors is emphasized, with prime attention being given to the function of the secret. Clinical examples are provided to illustrate some issues of practice and technique. It is suggested that analytic treatment can proceed without strict adherence to the rule of 'no secrets'. Indeed, there may be times when it would be best not to seek the patient's secret, e.g. when the secret helps to organize a tenuous self-identity. Finally, there is discussion of some of the difficulties involved in being the receiver of patients' secrets.