Random assignment of participants to experimental and comparison treatments is believed to enhance the comparability of the study groups on baseline characteristics. Despite its benefits, random assignment presents threats to validity. It ignores participants' treatment preferences. If not accounted for when participants are allocated to treatments, preferences influence enrolment in the study, representativeness of the accrued sample, attrition, adherence to treatment, and outcomes. This methodological article describes the mechanisms underlying the influence of treatment preferences on the external and internal validity of an intervention evaluation study. The authors present empirical evidence to support the points of discussion. They describe alternative research designs that account for treatment preferences, for use in future nursing intervention research.