Childhood stroke is a severe but rare disorder with increasing incidence. Early research assumed an improved outcome due to better neuronal plasticity in childhood stroke patients as compared to adults. However, more recent studies indicate similar rates of disabilities and cognitive impairment. Here, we used resting-state fMRI to study transient brain states and their association with cognitive outcome in children and adolescents after childhood stroke. We used clustering analysis to characterize eight recurring spatial co-activation patterns (CAPs) in a cohort of 19 patients and 17 age-matched controls. For each individual, we quantified CAP-related occurrence rates and dwell times, as well as transition probabilities between CAPs. Most pronounced differences between patients and controls were observed for CAPs delineating the main parts of the default mode network (DMN). Specifically, transient states of DMN deactivation were less occurring in the patient group. Moreover, we observed that transitions from DMN-related brain states were associated with cognitive performance in patients but not in controls. This study is the first to investigate the impact of default- and active modes on cognitive performance. Our findings suggest that the ability to switch between a default mode and an active mode might be an early predictor for cognitive outcome in patients following childhood stroke. Overall, or results emphasize the significance of flexible temporal transitions between brain states for cognition.