In this article, we examine whether the effectiveness of processing instruction (PI) is limited to forms targeted in the instructional treatment (primary effects) or whether it also extends to other forms (transfer-of-training effects). L2 Spanish learners (N = 123) received either PI or traditional instruction (TI) targeting third-person accusative clitic pronouns, and a third group received no instruction. We used a pretest, immediate posttest, and delayed posttest design to examine the impact of instruction type on learners' interpretation and production of accusative clitic pronouns; we used the same design to analyze the impact of instruction type on their interpretation and production of a second form, dative clitics, for which learners did not receive instruction (secondary effects). Although both PI and TI improved on the learners' interpretation of dative clitics in object-verb-subject (OVS) sentences, TI showed a decrease in accuracy with subject-verb-object (SVO) sentences. We propose that of the two approaches, only PI is effective in prompting learners to alter the way they process primary linguistic data in the input.