This article presents results from a recent study that isolated grammar instruction that is deductive (i.e., involving rule presentation and metalinguistic information) as a variable and contrasted it with an instructional treatment that is inductive (i.e., focusing on form with no explicit grammar instruction). The effectiveness of these two types of instruction was compared on measures of both comprehension and production. The study also investigated the interaction between type of instruction and the morphological and syntactical features involved in the acquisition of direct object pronouns in French as a second language. The results revealed a significant advantage for the deductive instruction group. The study highlighted the difficulty of designing language measures that access implicit language knowledge. It also underlined the strong relationship that exists between the observed effectiveness of a particular type of instruction and tests/measures used.