For the development of phonetic rules in an allophone based text-to-speech system a strategy has been adopted which consists of three major elements, viz. (l) a tool that allows to test new rules without having to go through the time consuming process of compiling and linking, (2) efficient tools for obtaining acoustic data to build rules on, and (3) a schedule for a series of formal intelligibility tests. It is explained how these three elements make up a procedure for speech synthesis rule development: An interactive parameter adjustment module enables the rule developer to test the relevancy of speech data obtained by means of efficient signal processing techniques and intelligibility tests, before they are converted into a (set of) rule(s). The intelligibility tests carried out so far indicate that the adopted approach is promising.