A total of 5473 pharmacological stimulation tests were carried out in 3143 children and subjected to statistical analysis. The mean chronological age of the children was 9 years 9 months (range 3 years to 16 years 6 months) and mean bone age was 7 years 6 months (range 2 years to 14 years). Nine pharmacological tests were used: (I)arginine(n = 625); (2) clonidine (n= 339); (3) insulin (n= 198); (4) ornithine (n =162); (5) insulin and arginine (n= 203); (6) clonidine and betaxolol (n =2003); (7) L‐dopa (n = 685); (8) glucagon and propranolol (n=443); and (9) glucagon and betaxolol (n = 815). Measurement of plasma growth hormone was always performed using the same method. The distribution of values in each test was of the gausso‐logarithmic type. The results of the mean peak and the 95% confidence limit were as follows: (1) 10.2, 0.45; (2) 11.5,0.7; (3) 11.8,0.8; (4) 14.2,1.2; (5) 14.3, 0.9; (6) 15.7, 1.1; (7) 19.8, 2.1; (8) 20.8, 2.3; (9) 21, 2.5. These results lead to the following conclusions: the specificity of these tests is low, the mean peak may vary two‐fold from one test to another, and the percentage of peaks<10 ng/ml ranges from 69% for test 1 to 29% for tests 8 and 9. The proportion of growth hormone deficiencies thus varies considerably according to the test used. To avoid these disparities, we propose a weighting index to adjust values according 40 the tests used: weighting index (1) 1.9; (2) 1.48; (3) 1.4; (4) 1.16; (5) 1.06; (6) 1.01; (7) 0.73; (8) 0.69; (9)