Self-Compassion Scale: 26 statements about how you treat yourself
Most self-report scales ask what you feel. The SCS asks what you do with the feeling once it arrives. Fail at something and one voice says of course you did. Another says this happens to everyone. The scale was built to measure which voice wins.
Kristin Neff published the 26-item version in 2003, built around three pairs: self-kindness against self-judgment, common humanity against isolation, and mindfulness against over-identification. She treats the six parts as one system rather than three separate traits.
The six subscales
Each pair is measured by two subscales, so a report shows six scores rather than three. That split matters, because the pairs are not simple opposites. Someone can be warm toward themselves in private and still believe their problems cut them off from everyone else.
| Subscale | Items | What a high score means |
|---|---|---|
| Self-Kindness | 5 | Warmth and support turned inward |
| Self-Judgment | 5 | Criticism and blame when something goes wrong |
| Common Humanity | 4 | Failures read as shared human experience |
| Isolation | 4 | Feeling cut off from others while struggling |
| Mindfulness | 4 | Steady awareness of painful thoughts |
| Over-Identification | 4 | Getting swept up and fused with them |
Each item is answered on a 5-point scale. The three negative subscales are reverse-scored, and the total is an average between 1 and 5 rather than a raw sum.
Where the boundaries come from, and where they do not
There is no clinical cutoff. Neff's information sheet offers descriptive ranges for the mean: 1.0 to 2.49 low, 2.5 to 3.5 moderate, 3.51 to 5.0 high. Those are labels for reading a number, not thresholds that divide one kind of person from another, and nothing in the literature establishes them as diagnostic.
One published sample averaged 18.25 (SD 3.75) on the 26-item sum. Read that as a reference point for how a general sample answered, not as a target to hit.
Reliability
Internal consistency for the total was .92. The subscales are thinner because they hold four or five items each: Self-Kindness .78, Self-Judgment .77, Common Humanity .80, Isolation .79, Mindfulness .75, Over-Identification .81. Three-week retest was .93 for the total and between .80 and .88 for the subscales.
What does a single number represent here? A bifactor analysis across 20 samples (N = 11,685) found one general factor carrying 95% of the reliable variance alongside six specific ones, which Neff reads as support for using the total. Measurement invariance has been tested in 18 countries. Muris and Otgaar argue the other side: that positive and negative self-compassion behave as separate systems and a total score hides the difference. The disagreement is unresolved. A total on this page is the version the author defends.
How it lines up with other measures
| Measure | r |
|---|---|
| Beck Depression Inventory | -.51 |
| Trait anxiety (STAI) | -.65 |
| Self-criticism (DEQ) | -.65 |
| Social connectedness | .41 |
| Satisfaction with Life Scale | .45 |
| Social desirability | .05 |
The negative correlations are moderate to strong, which is roughly what a protective construct should look like. The near-zero social desirability correlation is the more useful number: people are not scoring high because they want to look good. Self-compassion is also independent of self-esteem and predicts outcomes beyond it.
Neff's 2023 review covers more than 3,800 publications. The practical thread running through them is that the skill moves. Mindful Self-Compassion training and compassion-focused therapy both shift the scores.
What this test is not
A high score is not self-indulgence, and the data does not support the idea that self-compassion makes people lazy. It tracks with higher intrinsic motivation and a stronger sense of responsibility. If your six scores sit low across the board, that describes a habit, not a character. A scale cannot diagnose anything, and this one was never built to try. This page is not a diagnosis and does not replace one.