Two experiments explored how learners allocate limited time across a set of relevant on-line texts, in order to determine the extent to which time allocation is sensitive to local task demands. The first experiment supported the idea that learners will spend more of their time reading easier texts when reading time is more limited; the second experiment showed that readers shift preference towards harder texts when their learning goals are more demanding. These phenomena evince an impressive capability of readers. Further, the experiments reveal that the most common method of time allocation is a version of satisficing (. Reader and Payne, 2007) in which preference for texts emerges without any explicit comparison of the texts (the longest time spent reading each text is on the first time that text is encountered). These experiments therefore offer further empirical confirmation for a method of time allocation that relies on monitoring on-line texts as they are read, and which is sensitive to learning goals, available time and text difficulty.