Purpose <p>Loneliness is a widely studied transdiagnostic risk factor for myriad mental and physical health outcomes. However, it is often treated as unidimensional, potentially obscuring complex patterns of health disparities across different manifestations. This study sought to (1) explore the granularity of loneliness and (2) examine potential disparities in loneliness dimensions across groups of youths defined by self-reported gender identities and sexual orientations.</p> Method <p>This secondary analysis uses baseline data from a web-based randomized controlled psychological intervention trial, for which 2,446 youths completed the UCLA Loneliness Scale. Bass-Ackwards analysis extracted loneliness dimensions and associated factor scores at different levels of resolution. Cross-group comparisons focused on the most comprehensive level that retained interpretability. ANOVAs and Tukey’s tests compared groups defined as (a) cisgender heterosexual, (b) cisgender sexual minority, and (c) dual gender and sexual minority.</p> Results <p>We identified a maximum of five interpretable loneliness dimensions: superficial relationships, relationship incompatibility, lack of support, aloneness, and introversion. Significant cross-group disparities emerged for superficial relationships, aloneness, and introversion, but not for relationship incompatibility or lack of support. Relative to cisgender heterosexual individuals, levels of superficial relationships and aloneness were disparately higher for sexual minorities and dual gender/sexual minorities, while levels of introversion were higher for dual gender/sexual minorities only.</p> Conclusion <p>Multidimensional conceptualizations of loneliness may reveal distinct epidemiological patterns across demographic groups that can be obscured by a unidimensional approach. Future work should examine whether the multidimensionality of psychosocial risk factors like loneliness may be pertinent for applied research.</p>