Prior evidence from masked morphological priming has revealed conflicting findings regarding the acquisition of morpho-orthographic segmentation mechanisms in developing readers. Here, we examined changes in masked morphological priming across grade within a large sample of French primary school children (n = 191, Grades 2-5) and how these effects are modulated by individual differences in reading proficiency, spelling proficiency, and morphological awareness. Target words were preceded by either (a) a suffixed word prime (e.g., tristesse-TRISTE), (b) a suffixed nonword prime (e.g., tristerie-TRISTE), (c) a non-suffixed nonword prime (e.g., tristald-TRISTE), or (d) an unrelated prime (e.g., direction-TRISTE) using very short prime durations (50 ms). Moreover, a frequency manipulation was included for suffixes and non-suffixes. The results revealed robust suffixed word priming across all children independent of grade and proficiency. On the other hand, priming in the suffixed and non-suffixed nonword conditions was modulated by reading proficiency, with high-proficiency children showing facilitation and low-proficiency children showing inhibition. The effects of suffix and non-suffix frequency were modulated by grade, with decreasing effects as grade increased. None of the observed priming effects were modulated by grade, spelling proficiency, or morphological awareness. The results suggest that reading proficiency is an important predictor for embedded stem activation mechanisms in primary school children, which we discuss in the context of recent theories of morphological processing.