Do children with better reading competence read more, or do avid readers increase their reading competence? This highly relevant question has been discussed for many years, yet conclusive results are rare. Previous studies suffer from small sample sizes or omitted variable bias, rendering their findings questionable. We provide new insight using large-scale German panel data (N > 5100). By surveying secondary school students (initial age = 11.1 years, SD = 0.50 years) in grades 5, 7, and 9, we can trace their reading competence and exposure over time. We estimate random intercept cross-lagged panel models (RI-CLPMs) to analyze time-dependent within-student changes and dynamic feedback loops. We include relevant control variables to account for potential confounding and conduct robustness checks using a classical CLPM. Our findings show no evidence that reading more results in better reading competence. However, higher reading competence may lead to increased reading exposure. Testing for gender differences reveals that these conclusions hold for both boys and girls and that no interactions with gender are detectable, despite girls tending to read more and having a higher reading competence on average.