This article presents a new model for the main hydration peaks of tricalcium silicate and alite on the assumption that C-S-H nucleates and grows as needles. The model relies only on directly observable quantities and reproduces the transition from the acceleration to deceleration periods without assuming a diffusion barrier or impingement. Simulations of the model based on measured characteristics of the grains and needles are compared with experimental data and satisfactory agreement is found across a wide range of experiments. Because the model disentangles the impact of each input parameter, it sheds new light on some aspects of alite hydration. In particular, the model provides a quantification of the impact of the consumption of the small grains and shows that it should not be neglected as soon as several tens of percent of the hydration reaction is reached. Finally, the article provides a critical examination of other models.