The present study investigated the role of overt phonological realisation of morphological marking on the implementation of subject-predicate agreement in language production. This study was conducted in Hebrew, and focused on subject-predicate gender agreement for inanimate nouns. In Hebrew, singular masculine forms are usually morphologically unmarked, whereas singular feminine forms are morphologically marked. Although the system is generally consistent, it includes examples of feminine forms which are not marked with feminine suffixes, i.e., unmarked feminine forms. Likewise, there are particular suffixes that usually denote masculine or feminine plurals. However, there are examples of masculine plurals which are inflected irregularly, with a feminine suffix. We took advantage of these characteristics of Hebrew to manipulate morphological factors in the process of agreement production. Using a sentence-completion task for complex noun phrases, we tested the effect of the absence of overt gender marking (Exp. 1) and misleading gender marking (Exp. 2) of the local noun on the frequency of agreement errors. The results revealed that whereas the absence of overt gender marking did not affect the frequency of agreement errors, overt misleading gender marking did affect it. The results are discussed in relation to the distinction between inflectional and derivational structures and their relevance to the process of computing agreement in the Marking and Morphing model (Eberhard, Cutting, & Bock, 2005). (Contains 5 tables and 13 footnotes.)