In the modern era, the prevailing model of public education has been that of “one size fits all”, with private schooling being a small but notable exception. Language (of instruction) was generally viewed as a minor variable readily overcome by standard classroom instruction. As researchers have sharpened their focus on the reasons for educational failure, language has begun to emerge as a significant variable in producing gains in educational efficiency. This paper reports the intermediate result of a controlled study in a very rural area of a developing country designed to examine the effect of language of instruction on educational outcomes. In the experimental schools, children are taught to read first in the local language (via the local language) and are taught other key subjects via the local language as well. English is taught as a subject. Teachers in the control or standard schools continue the standard national practice of teaching all subjects in either English or Filipino, neither of which is spoken by children when they begin school. Year-end standardised testing was done in all subjects throughout grades one to three as a means of comparing the two programme methodologies.