In this paper we describe a strategy to estimate by CE the acidity constants (pKa) of complex polyprotic peptides from their building peptide fragments. CE has been used for the determination of the pKas of five short polyprotic peptides that cover all the sequence of amyloid beta (Aβ) peptides 1-40 and 1-42 (Aβ fragments 1-15, 10-20, 20-29, 25-35 and 33-42). First, the electrophoretic mobility (me) was measured as a function of pH of the background electrolyte (BGE) in the pH range 2-12 (bare fused silica capillary, I=25mM and T=25ºC). Second, the mes were fitted to equations modelling the ionisable behaviour of the different fragments as a function of pH to determine their pKas. The accuracy of the pKas was demonstrated predicting the electrophoretic behaviour of the studied fragments using the classical semiempirical relationships between me and peptide charge-to-mass ratio (me vs. q/Mr1/2, classical polymer model, q=charge and Mr=relative molecular mass). Separation selectivity in a mixture of the fragments as a function of pH was evaluated, taking into account the influence of the EOF at each pH value, and a method for the simple and rapid simulation of the electropherograms at the optimum separation pH was described. Finally, the pKas of the fragments were used to estimate the pKas of the Aβ peptides 1-40 and 1-42 (tC and D 3.1, E 4.6 and Y 10.8 for acidic amino acids and tN-D 8.6, H 6.0, K 10.6 and R 12.5 for basic amino acids), which were used to predict their behaviour and simulate their electropherograms with excellent results. However, as expected due to the very small differences on q/Mr1/2 values, separation resolution of their mixtures was poor over the whole pH range. The use of poly(vinyl alcohol) (PVA) coated capillaries allowed reducing the electroosmotic flow (EOF) and a slight improvement of resolution.