Objectives <p>To offer another lens to study how mindfulness influences behaviours and social relationships, this paper reports the creation of the Observed Mindful Behaviours (OMB) scale. The OMB responds to limitations in current evidence in mindfulness research, including the reliance on self-report data.</p> Method <p>Refinements to an existing 9-item scale were tested in two samples (<i>n</i> = 99, <i>n</i> = 101) using item response theory and confirmatory factor analysis. Survey data from 190 dyads (<i>N</i> = 380) were used to test construct validity of the refined scale using correlations and regression models within a proposed nomological network for observed mindful behaviours.</p> Results <p>A three-dimensional hierarchical model was confirmed for the 9-item OMB (SRMSR = 0.044, RMSEA = 0.070, CFI = 0.97, <i>w</i>_<i>h</i> = 0.90). Good alignment with trait mindfulness (<i>β</i> = 0.42, <i>R</i>² = 0.15) and interpersonal mindfulness (<i>β</i> = 0.17, <i>R</i>² = 0.12) supported criterion validity. The construct assessed by the OMB significantly converged with empathy (<i>r</i> = 0.19) and psychological capital (<i>r</i> = 0.22) and significantly diverged from psychological inflexibility (<i>r</i> = −0.31), distress (<i>r</i> = −0.29), and anger reactivity (<i>r</i> = −0.26), but not with prosocial intentions.</p> Conclusions <p>The OMB scale supersedes the Observed Mindfulness Measure. The scale detects the extent to which a person known to the rater (family, friend, or colleague) behaves in a way that is noticeably attentive, aware, and accepting (or mindful). Differential alignments with behavioural drivers (empathy, acceptance, psychological capital) and with behavioural states (distress, anger) help clarify what the OMB assesses. The OMB can be used to triangulate and strengthen self-reported findings, and the subscale scores can advance study into how mindfulness comes across to others.</p>