Because of the oscillatory pattern of upper airway resistance and breathing during sleep in patients with obstructive sleep apnea (OSA), we hypothesized that OSA patients have an underlying instability of ventilatory drive to inspiratory muscles. To assess the stability of ventilatory drive in OSA patients and controls, we used the pseudorandom binary stimulation (PRBS) test and examined the closed- and open-loop responses to hyperoxic hypercapnia. The closed-loop response is produced by interactions of dynamic gain in controller, plant, and ventilatory feedback. The open-loop response reflects controller dynamic gain or frequency-dependent chemosensitivity. As compared with 16 nonapneic, nonobese control subjects, a group of nine obese OSA patients had a higher peak response and a more rapid and irregular recovery phase of the closed-loop CO2 response in the PRBS test. The two groups had similar open-loop responses in the PRBS test, suggesting that central dynamic CO2 chemosensitivity was not abnormal in OSA. We conclude that the differences between OSA patients and controls in the closed-loop response in the PRBS test are not due to differences in dynamic controller gain, but are related to differences in dynamic plant gain and/or negative ventilatory feedback. In addition to OSA, obesity may affect these variables and may have been responsible for our findings.