Objectives <p>This study aimed to translate and validate the State Self-Compassion Scale in its long (SSCS-L, 18 items) and short form (SSCS-S, 6 items) for German-speaking samples and to investigate its associations with psychological well-being and mental health.</p> Method <p>An online sample (<i>n</i> = 1,436) completed the translated SSCS-L and other psychological state and trait measures. Factor structures were examined using Exploratory Structural Equation Modeling (ESEM). Associations between subscales of SSCS-L and other constructs were investigated using partial correlational network models.</p> Results <p>A 6-factor ESEM based on 16 items showed the best fit for the SSCS-L; a global self-compassion factor—and thus using a total score—was not supported. Subscales self-kindness and self-judgment showed acceptable to good internal consistency, all others only marginally acceptable or fair internal consistency. With the SSCS-S, a 2-factor ESEM fits best, representing positive compassionate and negative non-compassionate self-responding. The network model showed positive unique links between positive subscales of SSCS-L and predictors and indicators of well-being; and negative unique links between negative subscales and these indicators. Negative subscales of SSCS-L were positively related to mental distress, while positive subscales showed inverse associations.</p> Conclusions <p>We present the 16-item SSCS-L and 6-item SSCS-S as useful tools for assessing state self-compassion as a multidimensional construct in research and interventions. We recommend using the SSCS-L with its six and the SSCS-S with its two subscales, and advise researchers to check factor structure and reliability in their samples due to potential variability across contexts.</p> Preregistration <p>The study was preregistered in PsychArchives (<a href="https://doi.org/10.23668/psycharchives.6665">https://doi.org/10.23668/psycharchives.6665</a>), with deviations reported in the Online Resources.</p>