Impaired respiratory muscle endurance (RME) could reduce exercise tolerance and contribute to ventilatory failure. The aim of the present study was to develop a clinically-feasible method to measure RME using negative-pressure inspiratory-threshold loading. It was hypothesized that endurance time (tlim) could be predicted by normalizing oesophageal pressure-time product (PTP) per total breath cycle (PTPoes) for maximum oesophageal pressure (Poes,max); the load/capacity ratio. The corresponding mouth pressures, PTPmouth and Pmouth,max were also measured. The RME test was performed on 30 healthy subjects exposed to the same target pressure (70% of Poes,max). Eight patients with systemic sclerosis/interstitial lung disease were studied to assess the validity and acceptability of the technique. Normal subjects showed a wide intersubject variation in tlim (coefficient of variation, 69%), with a linear relationship demonstrated between log tlim and PTPoes/Poes,max (r=0.88). All patients with systemic sclerosis/interstitial lung disease had normal respiratory muscle strength, but six out of eight had a reduction in RME. In conclusion, endurance time can be predicted from the load/capacity ratio, over a range of breathing strategies; this relationship allows abnormal respiratory muscle endurance to be detected in patients. Oesophageal and mouth pressure showed a close correlation, thus suggesting that the test could be applied noninvasively.