Increasing use of the coefficient of permeability as a quality control parameter for earthworks, in particular for landfilling, has resulted in many engineers looking to predict coefficientsof permeability for clays from more widely available but often widely scattered data. In a number of cases this has led erroneously to the interpretation of the moisture content coefficient of permeability relationship as the direct inverse of the moisture content-density relationship. This paper re-examines the original theories proposed for movement of water through soils of differing grain size and goes on to show that extrapolation of the permeability-density relationship for granular soils to finer grained soils is incorrect. Laboratory permeability tests were conducted on four different soil types using the falling head permeameter to investigate the relationships between permeability, dry density and moulding moisture content for each soil. Coefficients of permeability were determined for moisture contents up to the liquid limit of the soil. It is shown that a minimum value of permeability is obtained close to optimum moisture content and that this is maintained to within sensible engineering limits for moisture contents up to that of the liquid limit. Results obtained for predominantly granular soils confirm that permeability is a function of density but, for more cohesive soils, permeability is unaffected by changes in moisture content wet of optimum. Conclusions emphasise that to achieve minimum possible coefficients of permeabilitythere is a need for soilwater content at placement to be at or slightly above optimum moisture content and for pelletization of clay material.