The aim of the study was to investigate the effect of spending one night without sleep on the performance of complex cognitive tasks, such as problem-solving, in comparison with a purely short-term memory task. One type of task investigated was immediate free recall, assumed to reflect the holding capacity of the working memory. The other type of task investigated was represented by syntactical reasoning and problem-solving tasks, assumed to reflect the processing (the mental transformation of input) and monitoring capacity of the working memory. Two experiments with a repeated-measures design were performed. Experiment 1 showed a significant decline in performance as a function of sleep loss on Raven's progressive matrices, a problem-solving task. No other main effect of sleep loss was found. Experiment 2 had a different order between tasks than Experiment 1 and the time without sleep was increased. A number-series induction task was also used in Experiment 2. A significant, negative effect of sleep loss in performance on Raven's progressive matrices was found in Experiment 2. The effects of sleep loss on the other tasks were nonsignificant. It is suggested that Raven's progressive-matrices task reflects the ability to monitor encoding operations (selective attention) and to monitor mental "computations".