Summary Although complaints of impaired daytime functioning are essential to the diagnosis of primary insomnia, objective evidence for cognitive dysfunction has been hard to establish. A prerequisite for understanding the neurocognitive consequences of primary insomnia is to establish task paradigms that robustly differentiate insomniacs from well‐sleeping subjects. We hypothesized that the decline in performance that typically occurs with an increasing cognitive demand would provide a more sensitive measure than performance on a single task version. The hypothesis was tested, first, by assessing the performance on two vigilance tasks with different cognitive demands in 25 elderly patients with primary insomnia and 13 healthy well‐sleeping age‐matched subjects. Secondly, we investigated the performance response to sleep therapy using a waiting‐list controlled design. Sleep therapy consisted of a multi‐component intervention including sleep restriction, cognitive behavioral therapy, bright‐light therapy, structured physical activity and body temperature manipulations. The results show that insomniacs differed markedly from controls in their reaction times across tasks with different cognitive demands: patients responded faster on the ‘simple’ vigilance task, yet slower on the ‘complex’ vigilance task. Sleep therapy effectively restored normal performance: patients became significantly slower on the ‘simple’ task and faster on the ‘complex’ task, returning to the performance levels of control subjects. These findings indicate that the performance decline associated with increasing cognitive demands is possibly the first sensitive and robust measure of the neurocognitive sequelae of insomnia. We suggest that future studies on cognition in primary insomnia should apply a design that varies task demands.