<p><b>Background:</b> Disability does not only depend on individuals' health conditions but also the contextual factors in which individuals live. Therefore, disability measurement scales need to be developed or adapted to the context. Bangladesh lacks any locally developed or validated scales to measure disabilities in adults with mobility impairment. We developed a new Locomotor Disability Scale (LDS) in a previous qualitative study. The present study developed a shorter version of the scale and explored its factorial structure.</p> <p><b>Methods:</b> We administered the LDS to 316 adults with mobility impairments, selected from outpatient and community-based settings of a rehabilitation centre in Bangladesh. We did exploratory factor analysis (EFA) to determine a shorter version of the LDS and explore its factorial structure.</p> <p><b>Results:</b> We retained 19 items from the original LDS following evaluation of response rate, floor/ceiling effects, inter-item correlations, and factor loadings in EFA. The Eigenvalues greater than one rule and the Scree test suggested a two-factor model of measuring locomotor disability (LD) in adults with mobility impairment. These two factors are 'mobility activity limitations' and 'functional activity limitations'. We named the higher order factor as 'locomotor disability'. This two-factor model explained over 68% of the total variance among the LD indicators. The reproduced correlation matrix indicated a good model fit with 14% non-redundant residuals with absolute values > 0.05. However, the Chi-square test indicated poor model fit (p < .001). The Bartlett's test of Sphericity confirmed patterned relationships amongst the LD indicators (p < .001). The Kaiser-Meyer-Olkin Measure (KMO) of sampling adequacy was .94 and the individual diagonal elements in the anti-correlation matrix were > .91.</p> <p>Among the retained 19 items, there was no correlation coefficient > .9 or a large number of correlation coefficients < .3. The communalities were high: between .495 and .882 with a mean of 0.684. As an evidence of convergent validity, we had all loadings above .5, except one. As an evidence of discriminant validity, we had no strong (> .3) cross loadings and the correlation between the two factors was .657. The 'mobility activity limitations' and 'functional activity limitations' sub-scales demonstrated excellent internal consistency (Cronbach's alpha were .954 and .937, respectively).</p> <p><b>Conclusions:</b> The 19-item LDS was found to be a reliable and valid scale to measure the latent constructs mobility activity limitations and functional activity limitations among adults with mobility impairments in outpatient and community-based settings in Bangladesh.</p>