We are exposed to an ongoing flow of information, without explicitly knowing temporal contingencies between successive events. Can we use such contingencies to predict information in time and space? Participants (N = 18) discriminated the spatial frequency of two gratings. Before the gratings appeared, two unattended cues were presented on the top and bottom of the screen with a 100 ms delay. In different blocks, unbeknownst to the participants, the temporal order of the cues predicted when or where stimuli appeared. Performance improved in the condition in which the temporal order of the cues predicted when stimuli appeared. In the second experiment (N = 18), we showed that the improvement was unlikely to have been caused by the spatial component of the temporal order between the cues. Thus, the temporal order of visual events can be exploited in the absence of explicit knowledge to orient attention in time.