A robust state estimation scheme is proposed for anaerobic digestion (AD) processes to estimate key variables under the most uncertain scenarios (namely, uncertainties on the process inputs and unknown reaction and specific growth rates). This scheme combines the use of the IWA Anaerobic Digestion Model No. 1 (ADM1), the interval observer theory and a minimum number of measurements to reconstruct the unmeasured process variables within guaranteed lower and upper bounds in which they evolve. The performance of this robust estimation scheme is evaluated via numerical simulations that are carried out under actual operating conditions. It is shown that under some structural and operational conditions, the proposed robust interval observer (RIO) has the property of remaining stable in the face of uncertain process inputs, badly known kinetics and load disturbances. It is also shown that the RIO is indeed a powerful tool for the estimation of biomass (composed of seven different species) from a minimum number of measurements in a system with a total of 32 variables from which 24 correspond to state variables.