We performed a two-minute incremental threshold loading test (incremental test) in ten normal subjects on three occasions, and having ascertained the maximum load (max load) against which they could inspire for two minutes, measured how long this load could be tolerated by these same subjects on three further occasions (tlim test). We compared the reproducibility of the two tests. There were no significant differences found in the mean max loads in the three incremental tests, or in the endurance times in the three tlim tests. However, the intraindividual coefficients of variation of max load in the incremental test (0 to 14 percent) were much smaller than the intraindividual coefficients of variation of endurance time in the tlim test (20 to 65 percent). We found that the large variability in endurance time in our tlim tests was most likely accounted for by variability in breathing pattern, inspiratory flow rate and breath-by-breath mouth pressure generation. Differences in these parameters did not, however, explain why in the tlim test a given subject could tolerate for 19 minutes a load only 100 g less than that which he was unable to tolerate for two minutes in the incremental test. These findings emphasize the differences between these two tests of respiratory muscle endurance. Since there was less intraindividual variability in the two-minute incremental threshold loading test, we suggest that this test may be more useful than the tlim test.