Measurement of dynamic cerebral autoregulation (CA), the transient response of cerebral blood flow (CBF) to changes in arterial blood pressure (ABP), has been performed with an index of autoregulation (ARI), related to the parameters of a second-order differential equation model, namely gain (K), damping factor (D) and time constant (T). Limitations of the ARI were addressed by increasing its numerical resolution and generalizing the parameter space. In 16 healthy subjects, recordings of ABP (Finapres) and CBF velocity (ultrasound Doppler) were performed at rest, before, during and after 5% CO2 breathing, and for six repeated thigh cuff maneuvers. The unconstrained model produced lower predictive error (p < 0.001) than the original model. Unconstrained parameters (K′-D′-T′) were significantly different from K-D-T but were still sensitive to different measurement conditions, such as the under-regulation induced by hypercapnia. The intra-subject variability of K′ was significantly lower than that of the ARI and this parameter did not show the unexpected occurrences of zero values as observed with the ARI and the classical value of K. These results suggest that K′ could be considered as a more stable and reliable index of dynamic autoregulation than ARI. Further studies are needed to validate this new index under different clinical conditions. © 2008 Institute of Physics and Engineering in Medicine.