The purpose of this study was to examine extensions of a model of hypoxic ventilatory decline (HVD) in humans. In the original model (model I) devised by R. Painter, S. Khamnei, and P. Robbins (J. Appl. Physiol. 74: 2007-2015, 1993), HVD is modeled entirely by a modulation of peripheral chemoreflex sensitivity. In the first extension (model II), a more complicated dynamic is used for the change in peripheral chemoreflex sensitivity. In the second extension (model III), HVD is modeled as a combination of both the mechanisms of Painter et al. and a component that is independent of peripheral chemoreflex sensitivity. In all cases, a parallel noise structure was incorporated to describe the stochastic properties of the ventilatory behavior to remove the correlation of the residuals. Data came from six subjects from a study by D.A. Bascom, J.J Pandit, I.D. Clement, and P.A. Robbins (Respir. Physiol. 88: 299-312, 1992). For model II, there was a significant improvement in fit for two out of six subjects. The reasons for this were not entirely clear. For model III, the fit was again significantly improved in two subjects, but in this case the subjects were those who had the most marked undershoot and recovery of ventilation at the relief of hypoxia. In these two subjects, the chemoreflex-independent component contributed approximately 50% to total HVD. In the other four subjects, the chemoreflex-independent component contributed approximately 10% to total HVD. It is concluded that in some subjects, but not in others, there may be a component of HVD that is independent of peripheral chemoreflex sensitivity.