The approach used to classify patients into "refuser" subgroups can influence conclusions about the relationship between refusal of antipsychotic medication and involvement in important hospital-based outcomes such as the need for seclusion and restraint. Use of a cross-sectional taxonomy led to conclusions which were somewhat negatively biased against "refusers." In contrast, use of a longitudinal taxonomy which reflects changes in formal informed consent behavior over time, suggests that it is not refusal, per se, but changes in informed consent status that appear to be associated with problematic behavior. In fact, "consistent-refusers" tended to be the least troublesome of the three informed consent status groups studied. Researchers are encouraged to use patient classification schemes which consider patterns of formal informed consent behavior over time when examining the relationship between refusal and involvement in important outcomes.