The present study used two letter detection tasks, the classic missing letter effect paradigm and a single word versus familiar word compound version of this paradigm, to study bottom-up and top-down processes involved in reading in normally achieving as compared to low achieving elementary school readers. The research participants were children in grades first to sixth who had been taught to read by three approaches to reading instruction (ARI): the whole language/global approach, the phonic/synthetic approach or the eclectic approach. Thus, the study attempted to clarify how different ARIs activate these processes in these two types of readers. The main hypothesis was that since low achieving readers rely on top-down processes for word recognition (see, e.g., Stanovich, 1980), the whole language ARI will reduce the difference in bottom-up tasks between them and the normally achieving readers. In the two experiments included in the study, participants were required to perform a letter detection task while reading short texts in Hebrew for comprehension. Contrary to this study hypotheses, the main finding was that the whole language ARI does not compensate for difficulties in bottom-up processes of low achieving readers. Moreover, the results of this study imply that any improvement in basic processes involved in reading proficiency produced by the whole language ARI for both the normal and the low achieving readers dissipates by grade three.