Two experiments investigated selective processing as a result of transient and sustained visual-spatial attention. In Experiment 1, attention was cued on a trial-by-trial basis and event-related brain potentials were measured to stimuli preceded by valid, invalid, or neutral symbolic precues. Trial validity had only small effects on posterior P1 and N1 components. At midline electrodes, an enhanced negativity for valid as compared with invalid trials was present, which appeared to reflect initial processing costs for invalid trials followed by an additional processing benefit for valid trials. Experiment 2 investigated whether these effects are specific for transient spatial attention by comparing transient and sustained attention conditions. No indication of early processing costs was found in the latter case.