Objectives: Interoception refers to the sensation, interpretation, and integration of internal somatic signals. Abnormalities in self-reported interoception are prevalent features of major depressive disorder (MDD) and may affect treatment outcomes. In the present study, we investigated the psychometric properties of the revised eight-dimensional and 37-item Multidimensional Assessment of Interoceptive Awareness questionnaire (the MAIA-2) in a severely depressed sample, after translating two updated scales (Not-Distracting, Not-Worrying) into German. Specifically, we examined the measure’s internal consistency reliability, sensitivity to change, and minimal important differences (MID) with a focus on patient’s antidepressive responses to treatment. Methods: The study enrolled 110 participants (age: M = 46.85, SD = 11.23; female: 55.45%) undergoing hospital treatment, of whom 87 were included in the pre-post analysis. Participants completed a German translation of MAIA-2 and the Beck Depression Inventory-II (pre-/post-treatment). Internal consistency reliability was determined by Cronbach’s α/McDonalds’s ω, sensitivity to change was determined by effect sizes, and MIDs were determined by distribution- (0.5*SD) and anchor-based approaches (mean change method; ROC curve cut-points). Results: Depression severity reduced over the course of treatment (Median = -65.22%), and 34.48% of patients achieved remission. Reliability was appropriate for post-treatment (range of ω: .70-.90), but questionable for two pre-treatment scales (Noticing: ω = .64; Not-Distracting: ω = .66). The eight dimensions of MAIA-2 were sensitive to change (standardized response mean: .32-.81; Cohen’s effect size: .30-.92). Distribution-based MIDs (.38-.61) and anchor-based mean change MIDs (remission vs. partial response: .00-.85; partial response vs. nonresponse: .08-.88) were established on the group level. For six scales, ROC cut-points (remission: .00–1.33; response: -.20–1.00) demonstrated accurate classification to treatment response groups on the individual level. Conclusions: This study demonstrated the applicability of the MAIA-2 questionnaire in MDD. The updated version may have led to reliability improvements regarding the revised scales, but subthreshold reliability was evident prior to treatment. The measure’s dimensions were sensitive to change. MIDs were established that corresponded with antidepressive treatment outcomes. Our findings are consistent with a growing area of research which considers somatic feelings as key contributors to mental health.