Maximal Inspiratory pressure (MIP) is an important clinical method used to assess respiratory muscle strength. The reliability and reproducibility of this measurement in mechanically ventilated patients is not certain. In 14 stable, mechanically ventilated patients, capable of spontaneous inspiratory efforts, we assessed maximal inspiratory efforts using the technique originally described by Marini and associates. MIP was measured in triplicate, by one to five experienced investigators, on one to seven successive days, for a total of 396 determinations on 54 patient days. The coefficients of variation among the triplicate efforts averaged 12 +/- 1%, indicating the test to be highly reproducible. There was significant variation among the MIP reported by different investigators studying the same patient on the same day (32 +/- 4%). The differences between best MIP by different investigators averaged 12.6 +/- 1.3cm H2O (40 +/- 4%). In 17 of 44 cases, one investigator placed MIP above -30cm H2O, whereas another placed it below. ANOVA showed that MIP was significantly affected by investigator (p less than 0.0001) as well as by patient (p less than 0.0001). Because "true" MIP must be equal to or greater than the best measured MIP, these data indicate that the MIP is commonly underestimated in patients receiving mechanical ventilation, even when standardized technique is used. Furthermore, our data show that reproducibility of triplicate MIP determination by a single observer does not indicate that the test is reliable.