Conventional characterisation of low alkalinity waters via pH measurement and titration of total alkalinity to a prescribed end-point invariably leads to large errors. These errors result from instability of the pH probe and an unknown titration end-point. In this paper two direct methods (termed the "double Gran function" and the "blend" method) for the characterisation of such waters are evaluated critically. A blend composed of the raw water, sodium chloride (to increase conductivity), and standard bicarbonate (to increase buffering capacity) was titrated with standard strong acid in two pH regions: 6.3 < pH < 7.0, and 3.5 > pH <4.0. In both methods, total alkalinity was determined using the latter set of points, and the first Gran Function. In the double Gran function method the upper set of titration points was used to determine CO2 acidity using the second Gran Function. In the "blend" method, equilibrium chemistry data were used to calculate total acidity for each point based on the known total alkalinity, pH reading, temperature ionic strength. The two methods gave excellent results (in terms of both repetition and accuracy) as compared to characterisation based on total alkalinity and inorganic carbon analysis. A detailed procedure for the execution of the two approaches is given in an appendix.