Borderline personality disorder has a lifetime prevalence of 5.9% in the largest epidemiological survey of US adults, and the DSM-5-TR gives a median point prevalence of 1.6%. On the Dark Mirror, the personality quiz that runs on this site, 419 of 6,507 takers (6.4%) came out with borderline as their leading axis, and 31 of them (0.48%) scored 60 or above on it.
Those are two different kinds of number, and this page keeps them apart on purpose. The research figures come from structured diagnostic interviews with tens of thousands of randomly sampled adults. Our figures describe the people who chose to sit a dark psychology quiz on a dark psychology creator's site, which is not the same population and never will be. Read them side by side. Do not read one as a check on the other.
Our numbers: the Dark Mirror Index, edition of 5 September 2026
Every figure in this section describes people who took this quiz. None of it is a population rate.
- The current edition has 6,507 takers. The site describes this publicly as a sample of 5,000 or more.
- 419 takers, or 6.4% of the sample, had borderline as their leading axis. Leading axis means the axis that collected more of their answers than any other, and nothing more than that.
- 31 takers, or 0.48% of the sample, scored 60 or above on the borderline axis. Sixty out of 100 requires 12 of the 20 answers to key to borderline. It is the top band on our reading scale, and it is a reading aid, not a diagnostic cut.
- The mean borderline score is 11.4 out of 100, which is 2.27 of 20 answers, with a standard deviation of 12.0 points. The average taker sits in the baseline band on this axis, and most of the variation sits below the traits line.
- The borderline axis has an internal consistency of KR-20 = 0.71, the second highest of the six axes. Only the neurotypical axis, at 0.87, holds together more tightly. That means the borderline questions are measuring something reasonably coherent within this instrument.
- Among the 419 borderline-leading takers, the most common second axis was neurotypical (214), then sociopathic (107), narcissistic (47), psychopathic (36) and histrionic (15). Half of the borderline-leading group leans neurotypical underneath.
- Across all 6,507 takers, borderline correlates with psychopathic at r = -0.415, a clear negative relationship.
- Against the other axes, borderline is close to flat: histrionic 0.066, neurotypical -0.118, narcissistic -0.119.
The published research
This section is other people's work, gathered by other methods, on other samples.
Grant and colleagues, publishing in the Journal of Clinical Psychiatry in 2008, put the lifetime prevalence of borderline personality disorder at 5.9%. That estimate comes from the second wave of the National Epidemiologic Survey on Alcohol and Related Conditions, a structured diagnostic interview administered to 34,653 US adults drawn as a national sample. The DSM-5-TR gives a median point prevalence of 1.6%, which is a different quantity: how many people meet criteria now, rather than at any point across a life. (Grant et al. 2008)
For contrast, the same survey programme put the lifetime prevalence of antisocial personality disorder at 3.6%, from the first wave and 43,093 adults. (Grant et al. 2004)
Those are diagnosis rates. They count people who cross a clinical bar on an instrument built to find them, administered by trained interviewers to a sample chosen at random. Nothing on this site does any of those three things.
The finding worth sitting with: borderline and psychopathic pull apart
The popular version of Cluster B treats it as one bundle. Borderline, narcissistic, antisocial and histrionic get talked about as neighbours who turn up to the same party, so people assume the traits travel together and that a high score on one predicts a high score on the next.
In this sample they do not. Borderline and psychopathic correlate at -0.415 across all 6,507 takers, and that is the strongest single relationship on our correlation board. As one rises, the other tends to fall. Meanwhile borderline sits near zero against histrionic (0.066), and only slightly negative against neurotypical (-0.118) and narcissistic (-0.119). So it is not that borderline is unrelated to everything. It is specifically, and fairly strongly, opposed to the psychopathic axis.
There is a plain mechanical reason to be careful here. The Dark Mirror has 20 questions and every answer is keyed to exactly one of six axes, so the axes compete for a fixed pool of answers. Any two axes will show some negative pull simply because an answer given to one is an answer denied to the other. That structural effect applies to every pair, though, and psychopathic is the only axis where borderline lands anywhere near -0.4. The other four sit between 0.066 and -0.119. Something beyond the arithmetic is separating these two.
The likelier reading is about what the two sets of questions ask for. The borderline items describe reactions that are felt too much and too fast: abandonment fear, a self that reshapes around whoever is in the room, a rage that arrives before the thought does. The psychopathic items describe the opposite failure mode: the flat response, the calculation still running while everyone else is feeling something. A person answering honestly is unlikely to sit at the top of both. The second-axis table says the same thing from another direction: of the 419 borderline-leading takers, only 36 had psychopathic underneath, while 214 had neurotypical.
Treat that as a finding about this instrument and this audience. It is not evidence about how the disorders co-occur in clinical populations, where comorbidity is measured by entirely different means and the answer is known to be messier.
How to read any of these numbers
Our figure will read higher than the clinical one, and that is worth explaining rather than burying.
The sample selects itself. People who take a dark psychology quiz on a dark psychology creator's site are not a community sample. They arrived because the subject already interests them, and a good share of them arrived because something in the subject already sounds like them. A survey run outside a gym finds that most people exercise. This is the same effect.
A screener is not an interview. Twenty self-report questions capture what somebody is willing to say about themselves in a few minutes. A diagnosis rests on history, duration, impairment, and a clinician's judgement about all three. The two instruments are not measuring the same object, so a 6.4% here and a 5.9% there are not rival estimates that happen to nearly agree. Their closeness is a coincidence of scale.
The bands are a reading aid. Baseline runs 0 to 15, traits 20 to 35, pronounced 40 to 55, and high is 60 or above. Those lines exist so that a number means something in a sentence. They are not clinical thresholds, and scoring in the high band on an axis is not evidence of a disorder. Nothing on this site diagnoses anybody, and nothing here tries to.
Read the Index as an audience portrait. Placed beside the literature it says something the literature cannot: this is how people who go looking at dark psychology describe themselves. That is a genuinely useful and genuinely different question. It is not a second opinion on prevalence.
How the numbers are made
The Dark Mirror asks 20 questions. Every answer is keyed to one of six axes: psychopathic, sociopathic, narcissistic, borderline, histrionic and neurotypical. An axis score is the count of answers keyed to it multiplied by five, so scores run from 0 to 100 in steps of five, and 60 means 12 answers out of 20.
For this edition, 6,526 takes were recorded. Four retakes by the same account and 15 takes finished in under 60 seconds were excluded, leaving the 6,507 takers reported above. A further 224 straight-line response sets were flagged in the published exclusion ledger. Reliability is reported per axis as KR-20, which is why the borderline figure of 0.71 appears above alongside the score distribution rather than instead of it.
The full instrument description, the exclusion ledger, the interval estimates and the axis correlations are on the methodology page. The live distributions, updated each morning, are on the Dark Mirror Index.
Related reading
For where borderline sits among its neighbours, read the Cluster B overview. For the presentation these statistics are least likely to catch, because it is the quietest, read the quiet borderline signs post. And for the distinction people get wrong most often, read borderline versus narcissistic.
Frequently Asked Questions
What percentage of people have borderline personality disorder?
Grant and colleagues, writing in the Journal of Clinical Psychiatry in 2008, reported a lifetime prevalence of 5.9% among 34,653 US adults interviewed for the second wave of NESARC. The DSM-5-TR gives a median point prevalence of 1.6%. Both figures come from structured diagnostic interviews with randomly sampled adults, not from online tests.
How many Dark Mirror quiz takers lead on the borderline axis?
Of the 6,507 people in the current edition of the Dark Mirror Index, 419 had borderline as their highest scoring axis, which is 6.4% of that sample. Only 31 of them, or 0.48%, scored 60 or above out of 100 on the borderline axis. That is a portrait of who takes this quiz, not a prevalence estimate for anybody else.
Does a high borderline score on a quiz mean I have BPD?
No. The Dark Mirror is a self-report screener with 20 questions, and its bands are a reading aid rather than a clinical threshold. A borderline personality disorder diagnosis is made by a clinician, over time, using a structured interview and a full history. A score tells you which way your answers leaned on one afternoon.
Do borderline and psychopathic traits travel together?
Not in this sample. Across all 6,507 takers the borderline axis and the psychopathic axis correlate at -0.415, meaning that as one rises the other tends to fall. That runs against the popular idea of Cluster B as a single bundle. It is a finding about this instrument and this audience, and it is not a claim about the disorders themselves.