The extended Kalman filter (EKF) is commonly used to filter out the inflow of noise into biological reactors. Its usefulness for bioreactors with monotonic outputs is well established. More recently, the EKF has been shown to be able to rescue stable periodic oscillations that have been distorted by noise. This study extends the use of the EKF to microbial oscillations that become chaotic under the influence of noise. As measured by the Lyapunov exponents of the noise-free and noise-filtered concentration profiles of a continuous culture of Saccharomyces cerevisiae, the filter is effective in recovering noise-free sustained oscillations from noise-induced chaos, but is less satisfactory for a culture with both deterministic and stochastic chaos. Other kinds of filters, employing artificial intelligence, are recommended in this case. © 2005 Elsevier B.V. All rights reserved.