The room acoustics treatments in conventional rectangular classrooms, if applied, usually appear as a sound-absorbing suspended ceiling in the entire ceiling. This approach to achieving a recommended reverberation time value underrates the importance of early reflections, particularly for speech intelligibility in the back rows. The research proposes a room acoustics design methodology named earlyreflection-oriented room acoustics design (ERORAD) for rooms for speech based on a model where early reflections are prioritised and quantified by the G50 parameter. On-site measurements were conducted in the "IYTE Yeni Amfi" room. A 3D computer model of the sample room was created to simulate and test various acoustic conditions. Binaural listening tests were conducted using the auralised audio material of the relevant scenarios mixed with background noise signals at varying levels representing the active classroom background noise. The study showed that an early-reflection-oriented room acoustics design (ERORAD) methodology increases relative sound levels of direct speech and its early reflections. This improves speech intelligibility at distant audience locations better than conventional approaches by categorising the room surfaces as functional surfaces (ERS) for early reflection of speech sound and as appropriate surfaces for absorption (SfA) for the absorption of the late reflections to control reverberation time. The findings suggested that G50 can be a primary parameter to determine the optimal trade-off point between speech sound energy and reverberation time to achieve required speech intelligibility in the audience positions away from the speaker at relatively high levels of active classroom background noise