We investigated the effects of age on proactive and reactive cognitive control in a large population sample of 809 individuals, ranging in age between 5 and 97 years. For that purpose, we used an anticue paradigm, which required a consistent remapping of cue location and response hand: Left-sided cues required right-hand responses and vice versa. After a random preparation interval of 100–850 ms, these anticues were followed by a target stimulus, which prompted a response with the index or middle finger of 1 of 2 hands. A neutral control condition involved uninformative cues, indicating all 4 possible response locations. The primary outcome measure was the difference between neutral and anticue reaction time (RT). Negative values indicated RT costs of the anticue, relative to the neutral condition, reflecting reactive cognitive control. Positive values indicated RT benefits, reflecting proactive cognitive control. Results were twofold. First, the switch from RT costs to benefits took place at longer preparation intervals in the youngest and oldest age groups than in the intermediate age groups. Second, irrespective of preparation interval, anticue performance followed an inverted U-shaped trajectory as a function of age, with a relatively steep improvement during childhood and adolescence, relative stability between 26 and 60 years, and a slightly accelerating decline into old age. Both patterns of results suggest an age-related transition from a primarily reactive, to a primarily proactive mode of cognitive control in early life and back again from a primarily proactive, to a primarily reactive mode of control in later life.