<p>Self-compassion interventions often combine cognitive and behavioral exercises, making it challenging to identify intervention targets that drive self-compassion, and ultimately, well-being. Our 1-week experiment examined whether cognitive vs. behavioral self-kindness uniquely impacts self-compassion and well-being and whether pre-registered moderators (gender, ethnicity, first-generation college status) influence intervention effects. Undergraduates (<i>N</i> = 739) were randomized into one of three conditions: cognitive self-kindness (thinking about oneself kindly), behavioral self-kindness (performing kind acts for oneself), or active control (listing daily activities). Participants completed their assigned activities three times over 1&#xa0;week. Participants in the behavioral self-kindness group reported greater posttest self-compassion compared to the cognitive self-kindness and control groups. Moderator analyses indicated that Latinx and first-generation college students uniquely benefited from exercising self-kindness—especially cognitive self-kindness. Being kind toward oneself may be a more effective short-term strategy to increase self-compassion compared to thinking kind thoughts about oneself, which might require more time and practice. Further research is needed to test cognitive self-compassion as an especially useful tool to bolster Latinx and first-generation college student well-being.</p>