Robust Eckart and Wiener filters used in time delay estimation have been considered previously. In the present paper results are extended and generalised to include two Hassab-Boucher filters, employed in the generalised crosscorrelator. These filters optimise two different criteria. Explicit solutions are derived for the ε-contaminated model and thep-point ε-contaminated model for the input signal and noise power spectral densities (PSDs). Theoretical results, numerical examples and computer simulations show that robust filters are less sensitive to the variations in the input PSDs and that they optimise the worst-case conditions.