Research has often emphasised that children who experience difficulties in learning to read get poorer in their performance, and the gap between good and poor readers increases over time which describes the well-accepted phenomenon called the Mathew Effect. However, some studies have shown that the Matthew Effect has not been confirmed. In this study, the reading growth curves of Turkish-speaking children and the existence of the Matthew Effect in their performances were examined. For this purpose, the reading skills of 482 children were measured twice a year in the first and second grades. A quadratic growth model successfully fit the data to explain overall reading performance across four-time points. Further, three latent classes, named as good, moderate, and poor readers, were identified on the basis of individual trajectories. Accordingly, the difference in performance between good readers and poor readers continued in all time points. Although the difference between classes continued to increase especially in the second and third time points, and supported the Matthew Effect, it showed a horizontal slope in the fourth time point. This slope shows that reading differences between classes may close in the future. The results were discussed on the basis of Turkish sample.