The ability to dynamically modulate the intensity of sustained attention (i.e., alertness) is an essential component of the human executive control system, allowing us to function purposefully in accordance with our goals. In this study we examine high-density ERP markers of alert responding during the fixed sequence sustained attention to response task (SART(fixed)). This paradigm has proven to be a sensitive clinical metric in patient populations with deficits in their ability to sustain attention (e.g., attention deficit hyperactivity disorder). In this task subjects withhold a button press to an infrequent no-go target ('3') embedded within a predictable sequence of numbers ('1' to '9'). Our data reveal a complex pattern of effects across the trial sequence of the SART, with clear contributions from frontal and parietal cortices to sustained attentional performance. Over occipito-parietal regions, early visual attention processes were increased during trial 2 (i.e., trial in which the digit '2' was presented) and trial 3, giving rise to the so-called selection negativity (SN). Two prominent late components were manifest during trial 2: LP1 (550-800 ms) and LP2 (850-1150 ms) over occipito-parietal and central sites. We interpret the LP1 component on trial 2 as reflecting retrieval of the task goal and the subsequent LP2 as reflecting competition between the currently relevant go response and the subsequent no-go response. On trial 3, an enhanced "no-go N2" (250-450 ms) was seen fronto-centrally in the absence of the "no-go P3" that typically follows. Fronto-polar activity was also seen across all trials and may be indicative of subgoal processes to integrate the association between stimulus and goal. Prior to a lapse of attention (i.e., failure to inhibit a response to "3") the LP1 was significantly attenuated on the preceding trial 2 indicating a failure of anticipatory goal-directed processing. The results are discussed in terms of models of sustained attention involving frontal and parietal cortices.