A model for predicting speech intelligibility of electro-acoustic listeners is introduced. The predicted SRTs are in the same range as the empirical SRTs with slightly different mean values. Increasing the electric field spread or internal noise variance increases predicted SRTs in electric-only and EA-listening. The EA-benefit seems to be influenced by the amount of complementary information provided by acoustic and electric feature. Increasing upper bound of residual hearing beyond 300 Hz did not change modeled SRTs in stationary noise for EA-listening.