BACKGROUND The diagnosis of growth hormone (GH) deficiency is based on a reduced peak GH response to provocative tests. However, the provocative tests are poorly reproducible and GH secretion is regulated by physiological parameters, such as body weight and puberty. The aim of this study was to assess the influence of body mass index (BMI) on GH response to provocative testing and to analyze the reproducibility of GH stimulation test. METHODS Clinical data were collected retrospectively by chart review from the Pediatric Endocrine Unit at the Ajou University Hospital. A total of 187 subjects with short stature who completed a GH stimulation testing between 2003 and 2009 were included in the study. RESULTS Of the 187 subjects, 66 (35.3%) had GH deficiency, while 121 (64.7%) were categorized as having idiopathic short stature. Reliability was calculated for 48 patients with idiopathic short stature who underwent the GH stimulation test twice. A GH response ≥10 ng/ml after retesting was found in 39 patients (81.3%) and a GH response <10 ng/ml was found in 9 patients (18.7%). In a stepwise multivariate analysis, BMI was a significantly independent predictor of peak GH. Elevated BMI was negatively associated with peak plasma GH levels. CONCLUSIONS The lack of reliability of GH values in response to pharmacological stimuli should be taken into account in the diagnosis of GH deficiency. Also, higher BMI is associated with lower GH secretion. BMI should be measured and GH results appropriately interpreted for all subjects undergoing GH stimulation testing.