This study aimed to investigate the clinical usefulness of an anthropometrically based method for estimating leg lean volume (LLV) in patients with chronic obstructive pulmonary disease (COPD) who presented or not with nutritional depletion. We prospectively evaluated a group of forty-eight patients (thirty-eight males) with moderate to severe COPD (Global Initiative for Chronic Obstructive Lung disease stages II-IV) who underwent a 6 min walking test and knee isokinetic dynamometry. Leg lean mass (muscle mass plus bone) was determined by dual-energy X-ray absorptiometry (DEXA) with derivation of its respective volume: these values were compared with those obtained by the truncated cones method first described by Jones and Pearson in 1969. As expected, depleted patients (n 19) had reduced exercise capacity and impaired muscle performance as compared to non-depleted subjects (P < 0.01). The mean bias of the LLV differences between anthropometry and DEXA were 0.40 litre (95 % CI - 0.59, 1.39) and 0.50 litre (95 % CI - 1.08, 2.08) for depleted and non-depleted patients, respectively. Anthropometrically and DEXA-based estimates correlated similarly with muscle functional attributes. A ROC curve analysis revealed that leg height-corrected LLV values had acceptable sensitivity and specificity to identify depleted patients (area under the curve 0.93 (range 0.86-1.00); P < 0.001). Moreover, patients with LLV <or= 9.2 litres/m (the best cut-off value according to the ROC curve) had significantly lower exercise capacity and muscle performance than their counterparts (P < 0.05). In conclusion, an anthropometrically based method of estimating LLV (Jones and Pearson method) was shown to present with clinically acceptable accuracy and external validity in depleted and non-depleted patients with stable COPD.