We evaluated event-related potentials (ERP) as indices of performance in three visual display-monitoring tasks: (a) signal detection, (b) running memory and (c) computation. Using factor analysis, we developed a global measure of performance (PF1) for each task. Task-relevant and irrelevant-probe stimuli elicited ERPs, which included components P1, N1, P2, P300, slow waves, and fronto-central negativities. In tasks (a) and (b), P300 amplitude in the task-relevant ERPs increased when the task was engaged, and was greater for accurate-than for inaccurate-response trials. In tasks (a) and (c), the irrelevant-probe ERPs also differed among task and performance conditions. To relate ERP measures to PF1, we developed linear regression models distinguished by three factors: general versus individual-subject, stimulus relevance, and signal-to-noise ratio (SNR). Model accuracy and reliability were highest for individual-subject, relevant-stimulus and high-SNR models, where average R 2 values for the three tasks were 0.44, 0.46, and 0.38, respectively. We discuss implications of the models for performance monitoring and implications of the ERP effects for human information processing.