Although snowball (chain referral) sampling has been widely used in qualitative sociological research, especially in the study of deviant behavior, the problems and techniques involved in its use have not been adequately. Specific problems areas include finding respondents and starting referral chains, and verifying the eligibility of potential respondents, engaging respondents as research assistants. Additional problems include controlling the types of chains and the number of cases in any chain and pacing and monitoring referral chains and data quality. The research described here aimed to understand how some people manage to break an addiction to heroin without any treatment, using samples of 100 people who had stopped using heroin on their own and a comparative group of 100 who quit after treatment. 30 references