Event-related potentials (ERPs) were recorded during a selective attention task involving weak or strong electrical stimuli delivered to the index fingers of the left and right hands. In an attend weak condition, subjects were asked to count the number of weak stimuli (targets) interspersed amongst strong stimuli (standards) delivered to a designated hand, whilst ignoring a similar set of stimuli delivered to the other hand. In an attend strong condition, subjects were asked to count the number of strong targets interspersed amongst weak stimuli. In both conditions, targets and standards occurred with probabilities of .10 and .40 respectively on each hand. Counting weak targets was found to be more difficult than counting strong targets. The latency of the earliest significant effect of selective attention on ERPs to standards was dependent on stimulus intensity: N80 in the case of weak standards, P105 for strong standards. There was no evidence of a later prolonged negative shift in attended standard ERPs. Rather, an enhanced N150 component post-centrally was followed by a prolonged positive shift of attended standard ERPs. This Late positive shift had a similar scalp distribution to the late positive component elicited by attended target stimuli.