Speaking tempo and intonation during Text to Speech Synthesis Sessions for the Greek language can be rendered with a set of duration rules and intonation models containing only the perceptually significant variability encountered in natural speech. Consistency of tempo over large synthetic texts may result to the enhancement of the listener's ability in comprehending the lexical message. This is achieved by eliminating any hesitations arising from a listener's effort to decode an erroneous tempo mismatch, the latter being often mistaken for a semantically meaningful tempo change. The achievement of consistency of tempo, therefore, contributes to the improvement of the fluency of synthetic speech. Severe degradation of the speech fluency in large sentences may also arise by improper registration of the intonation model, because any break of the target model at the sentence level may confuse the listener as being perceptually significant. The proposed intonation patterns along with their application rules, assure intonation consistency over sentences of any length, by assigning a perceptually significant intonation contribution to each individual lexical component. The proposed set of duration rules and intonation models is not only small but it also serves as a language specific component that is independent of the underlying synthesis strategy.