Our language environment is rich in regular patterns. It is therefore natural to assume that statistical learning—the ability to implicitly extract regularities from sensory input—plays a role in processing written text and contributes to more efficient reading. Given the moderate correlations between statistical learning performance and reading abilities found in the literature, we hypothesize that this relationship is mediated by individuals' sensitivity to orthographic regularities, such as frequent bigrams. Importantly, previous studies confounded sensitivity to bigram frequency with bigram legality and evaluations of reliability are lacking. In this presentation, I will discuss recent findings from multiple behavioral tasks designed to tap sensitivity to bigram frequency in adult Dutch and French speakers. These findings point to moderate reliability and consistently show near-zero or even negative correlations between sensitivity measures and reading abilities, suggesting that individual differences in bigram-level statistical learning do not contribute to reading abilities.