A general nmltlvanate method for determuung pH gradients in flow-mjechon analysis @IA) systems was developed An FIA system was constructed m which an inJected sample IS merged Hnth a non-lmear pH gradient and the method was demonstrated by the m]ectron of acid-base indicators ~nth pK, values spanning the range 3 4-8 9 The mdlcator responses were momtored by a photodiode-array spectrophotometer and by appbcation of the evolvmg factor analysis multivariate peak resolvmg technique it was possible to determme the concentration profiles of the acidic and basic forms of the indicators With this knowledge, the pH gradlent can be determined It 1s further shown that the pH gradient can be exploited to predxt pK, values of unknown indicators The predichon errors obtamed are comparable to literature results As an alternative, a multlvanate calibration model (partml least-squares regresslon) was tested for empincally predicting the pK, values by using the EFA concentration profiles as vanables The prediction errors obtamed by thus multivanate approach were also slrmlar to hterature results Keywords Flow system, Acidity constants, Evolvmg factor analysis, Indicators, Multwanate methods, pH gradients