BACKGROUND Incremental threshold loading (ITL) is a test of inspiratory muscle performance which is usually performed by breathing through a weighted inspiratory plunger, the load on the inspiratory muscles being increased by externally adding weights to the intake valve. This is not a true threshold device and may be inaccurate. This method was compared with a true threshold device consisting of a solenoid valve which only opens to supply air at a predetermined negative mouth pressure. METHODS Six naive, normal subjects (three men and three women) aged 22-24 years underwent three tests using each system. The inspiratory loads were increased every minute by equivalent amounts, -10 cm H2O with the solenoid valve and by 50 g with the weighted plunger, until the subjects could not inspire or sustain inspiration for a full minute. Six experienced subjects (four men and two women) aged 23-41 years were subsequently randomised to perform ITL with the solenoid valve, twice with the breathing pattern fixed and twice free. RESULTS The solenoid valve generated a more accurate mouth pressure response and was less variable at higher loads than the weighted plunger. The work performed (expressed as the pressure-time product) was less with the solenoid valve but was more reproducible. ITL with the solenoid valve was not influenced by controlling the breathing pattern of the subjects. CONCLUSIONS The solenoid valve has several features that make it superior to the weighted plunger as a device for ITL. It generates a more accurate mouth pressure response which is less variable at higher loads. Increases in load are smoother and quicker to introduce. ITL with the solenoid valve is not influenced by varying breathing patterns and does not require any external regulation.