The study explores two alternative explanations for the low attainment of Arabic-speaking students in reading literacy compared with Hebrew-speaking students in PIRLS (2006), i.e., one that associates variability in reading literacy scores with variability in socioeconomic measures, and another that considers the diglossia typical of Arabic (two linguistic codes: standard written Arabic and spoken form) as the critical variable responsible for the low attainment in reading literacy among Arabic-speaking students. Analyses of covariance were used to adjust achievement scores while statistically controlling for the effect of socioeconomic factors and consequently on the gap between the two groups. This resulted in a reduced achievement gap indicating the role of socioeconomic factors in explaining this gap. Repeating the analysis on mathematics and science scores, considered to be less affected by diglossia, enabled to compare the effect socioeconomic factors have on achievement in these subjects. Despite evidence on socioeconomic inequality between the Arabic and the Hebrew sector, measures of this inequality were found to have only a limited impact on reading achievement, but substantial impact in mathematics and science. The findings thus recommends educational interventions aimed to directly treat the problem of diglossia rather than focus only on improving socioeconomic conditions in schools.