We utilized PISA-2018 data from 71 countries to investigate the relationship between cognitive constructs employed in reading literacy (locating information, understanding, evaluating and reflecting, single and multiple thinking) and math and science achievement. We found that these cognitive constructs collectively accounted for 56% of the variance in math and 63% in science achievement, even after adjusting for gender, socioeconomic status, and country fixed effects. This means that the majority of cultural differences in math and science achievement (resulting from different education systems) can be explained by cognitive constructs employed in reading. We also noted that, at the country level, coefficients of cognitive constructs employed in reading demonstrated an interesting reconciliatory pattern. Countries with a weaker coefficient on the “locating information” dimension tended to have stronger coefficients on the “understanding” and “evaluation and reflection” dimensions; whereas, countries with a stronger coefficient on “understanding” dimension tended to have a weaker coefficient on “evaluation and reflection” dimension. These findings are particularly significant for STEM interventions aiming to enhance math and science achievement, as they indicate that a substantial portion of the variance in these achievements can be explained by cognitive constructs employed in reading literacy. Furthermore, culture-specific reconciliatory patterns imply that strengths in certain cognitive skills can compensate for weaknesses of others. Therefore, schools should consider modifying their curricula to integrate cognitive constructs employed in reading literacy more into math and science education.