Two groups were formed in a randomized, controlled trial: an Experimental group (n = 346), which benefited from specific interventions by trained teachers for a period of two years, and a Control group (n = 1354) with business as usual. For Experimental group in Kindergarten, the interventions were conducted in small groups and focused on code-related and language skills (mainly in comprehension), whereas in Grade 1, only code-related skills were addressed and the children received an amount of exposure that was a function of their performance level. Globally, Experimental group significantly outperformed Control group in code-related scores (Code) as well as on two reading scores (fluency and comprehension) at the end of Grade 1. Results were subdivided into quintiles for analysis as a function of initial performance. The main results showed that in Code Experimental group in the lowest quintiles obtained better performance at the end of Grade 1.