The cortisol awakening response (CAR) has been widely assessed as a measure of hypothalamic-pituitary-adrenal (HPA) axis activity. Short-term stability is high; however, little is known about the long-term stability of the CAR. Because there are indications that development in adolescence influences HPA axis activity, this study investigated the stability of the CAR over adolescence. Participants were 229 boys and 181 girls from an adolescent general population sample who were assessed in three consecutive years, at mean ages of 15.0 (SD=0.4), 16.0 (SD=0.4) and 17.0 (SD=0.4) years. Cortisol was analyzed in saliva sampled at awakening, and 30 and 60min later. Stability was investigated both as rank-order and as mean-level stability. Effects of physical development during adolescence on stability were investigated as well. Rank-order stability was moderate to low, with tracking coefficients (interpretable as stability coefficients over time) of .15 (p<.001) for cortisol at awakening and .24 (p<.001) for cortisol 30 and 60min after awakening. Mean-levels of cortisol at awakening did not change, while the response to awakening increased over the years (linear slopes for cortisol 30 and 60min after awakening all p<.01). The increase may reflect the physical development of the adolescents. This is the first study, in a large population based sample, indicating that the rank-order of the CAR is stable over the course of several years. Interestingly, mean-levels of the cortisol response to awakening increased over the years, suggesting a maturation of HPA axis reactivity in relation to physical development over adolescence. Physical development should therefore be taken into account when investigating the CAR as a measure of HPA axis activity in adolescence.