This quantitative survey study was designed to determine whether superintendent accountability behaviors or agreement about accountability behaviors between superintendents and their subordinate central office administrators predicted district student achievement. Hierarchical multiple regression and analyses of covariance were employed, controlling for district socioeconomic status. Superintendents who agreed with subordinate administrators about their accountability behaviors or underestimated these behaviors led districts with significantly higher student achievement in mathematics and reading than did superintendents who overestimated their accountability behaviors. Effect sizes were strong, with the self-other rating agreement category explaining 22% of the variance in mathematics scores and 19% of the variance in reading scores.