Abstract
Sampling opiate addicts outside the treatment system (hidden population) is considered difficult because there is no appropriate sampling frame. Several methods to achieve some form of random selection have been reviewed by earlier authors. One of these is 'snowballing' or snowball sampling. This term, however, covers various methods based on the central concept of locating new respondents through introductions from initial contact persons. In this publication theoretical aspects of one particular type of snowball sampling are described: snowballing with random nominee selection. Experiences with applying this type of snowballing in a medium sized Dutch city will be reported. The characteristics of the resulting study group of opiate addicts outside the treatment system are described together with different kinds of bias that may affect the representativeness of the study group.