Objective The current study aimed to identify the most statistically appropriate short‐form of the UCLA‐Loneliness Scale (LS). This was intended to provide researchers and practitioners with a measure that reduces the burden on participants and measures a unidimensional model of loneliness in accordance with the theory proposed by Russell. Method Eight short‐forms of the UCLA‐LS were assessed in two undergraduate samples (N = 552, 206). Students were from a metropolitan Australian university and completed online surveys. Results One‐, two‐, and three‐factor (i.e., an overall loneliness factor, and two method factors representing positively and negatively scored items) models of the original 20‐item UCLA‐LS provided poor model fit. One 10‐item short‐form provided adequate model fit in both samples. However, all configurations of the measure were generally internally consistent and displayed convergent validity. Greater loneliness was significantly associated with reduced social wellbeing, positive affect, life satisfaction, vitality, and psychological wellbeing, and increased negative affect. Conclusions It is recommended that a 10‐item short‐form is used in future applications. This short‐form displayed similar internal consistency and convergent validity as the 20‐item measure despite being half as long, and provided superior model fit.