Empirical research is needed to verify that the DSM-5 proposed diagnostic criteria for non-suicidal self-injury (NSSI) disorder provide a valid and clinically meaningful symptom set. The current study used data-driven methods to examine the diagnostic validity of the frequency and recency thresholds (i.e., Criterion A) for NSSI disorder. Data were collected from a random sample (n = 2206) of undergraduates. Participants completed counter-balanced questionnaires assessing the frequency, recency, and functions of NSSI, psychopathology symptoms, and indicators of distress and functional impairment. Discriminant functional analyses identified a significant differentiation for frequency between 10 or more acts of NSSI and 1–9 acts. Groups were also differentiated with a split on recency between less than 12-months ago and more than 12-months ago. After re-grouping the sample into categories based on the new frequency and 12-month recency threshold, the 10+ NSSI group reported significantly more functions for NSSI, including higher scores on affect regulation, self-punishment, and sensation seeking, than the subthreshold NSSI group. The 10+ NSSI group also reported significantly worse psychopathology, greater distress, and more impairment than both the subthreshold group and controls (no NSSI history). These findings indicate the current DSM-5 diagnostic criteria for NSSI disorder may be too liberal, and offer support for increasing the frequency thresholds for the diagnosis to ensure clinical validity and utility.