This study used longitudinal data from the Young Lives Project in Ethiopia to examine the main factors relating to preschool access and their potential effects on cognitive performance of children aged five and eight years. The results show that only one quarter of the preschool-aged children have the opportunity to attend this vital stage of education, with significant disparities by family wealth, education and regional location. Regardless of its limited coverage, preschool attendance is shown to have statistically significant positive effects on cognitive performance, measured by receptive vocabulary and mathematics tests. The effects do not also seem to fade away at a later age, as the inequality in cognitive abilities at age five continues to exist at the age of eight. Furthermore, using mediation analyses, causal chains between family backgrounds and cognitive performance were thoroughly analyzed. Bootstrap results show that preschool attendance mediates about one third of the direct effects of family wealth, education and regional location on child cognitive performance. Nevertheless, despite the importance of preschool education, public investment in this area is currently very limited, with the private sector taking the key role and exacerbating the inequality that exists between children of the rich and poor. These findings thus emphasize the need for government involvement in the form of public investment to this subsector to increase access for all children and reduce future educational inequalities. (author abstract)