External constraints on random sampling remains a constant challenge to nurse researchers seeking control over patient recruitment when investigating the outcome of nursing treatments. Traditional non-random sampling strategies employed in naturalistic clinical settings are prone to extraneous variables that frequently compromise the internal validity of the research design and confidence in the results. This paper explores the value of the Buglass scale and minimization in the context of a quasi-experimental research study to examine the management of deliberate self-ham by community psychiatric nurses. Discussion centres on their scope for control of extraneous variables without sacrificing the naturalistic setting within which community psychiatric nurses operate. A worked example illustrates the sampling procedure for minimizing differences between subjects assigned to a treatment and comparison group.