This paper reports an experiment that addresses whether manipulating foreknowledge of output tasks leads to differences in reading behaviour, text comprehension and noticing of language form. Three tasks are used: reading for memorization, reading for retelling and reading for visualization. Reading for memorization and reading for retelling are different kinds of FL (foreign language) output tasks, while reading for visualization does not involve any FL output. The participants' reading behaviour, text comprehension and noticing of language form were analysed using self-reports, comprehension test scores and retention or appropriate use of verb forms in the subsequent text production. The results suggest that foreknowledge of output tasks did influence reading behaviour: Participants used more translation into their L1, showed more matching between their interlanguage and target language, and paid more attention toward language form in the input. In addition, distinct reading behaviour and noticing of form were induced by different kinds of output tasks. Little difference was found in the text comprehension among the three groups. Implications from this study are (1) that language processing is affected by foreknowledge of output tasks--whether or not those output tasks are actually performed--and (2) different kinds of output tasks seem to cause different effects on language processing. (Contains 1 figure and 3 tables.)