The ACTH (adrenocorticotropic hormone) stimulation test is the gold standard for the diagnosis of adrenal insufficiency (AI), performed with ACTH high dose (HDT) or low dose (LDT). As salivary cortisol has been proposed as an alternative to serum cortisol, our aim was to evaluate the reliability of salivary cortisol compared to serum cortisol for diagnosing AI in children. Data were collected retrospectively. Salivary and serum cortisol values derived by 80 ACTH stimulation tests were obtained (39 F, 36 M; median age 11.5 years, IQR 6.9). Sampling was performed at baseline and after 30 and 60 min from ACTH administration during the HDT, and at baseline and 10, 20, 30, 40 and 60 min after the stimulation for the LDT. A serum cortisol level > 420 nmol/L ruled out AI. The correlation coefficients between serum and salivary cortisol for the HDT ( = 24) were 0.80 at t0, 0.48 at t30 and 0.75 at t60. All patients were adrenal sufficient. In 41% of the LDT, peak serum cortisol indicated insufficient adrenal function. The correlation coefficients between serum and salivary cortisol were 0.59 at t0 and 0.33 at the peak. For a cut-off of salivary cortisol < 15 nmol/L, sensitivity was 73.9% and specificity 69.6%. Our data do not support salivary cortisol as a valid alternative to serum cortisol during LDT. Regarding the HDT, results are more encouraging, however, further studies are needed.