OBJECTIVE The diagnosis of growth hormone (GH) deficiency (GHD) is routinely based on the results of two stimulating tests performed with different stimuli. Arbitrarily established, equal cutoff levels for the tests with different pharmacological stimuli, as well as a relatively high incidence of falsely decreased (FD) response in the tests have been reported. Falsely decreased GH response in one of the two performed tests does not exclude FD response in the second one, so, it seems very important to assess the incidence of FD response in both GH stimulating tests. METHODS The analysis comprised the results of two GH stimulating tests (the 1st one with clonidine, the 2nd one with either insulin or glucagon) performed in 780 short children. RESULTS The mean GH peak in the test with clonidine was significantly higher than that in both other tests (p<0.01). The rate of decreased GH secretion was 48.7 % in the test with clonidine, 80.5 % in the test with insulin and 81.5 % in the test with glucagon. Similar frequency of normal and subnormal test results was obtained for the cut-off value for the tests with insulin and with glucagon at the level of 6.4 ng/ml. The correlation between the results of the two tests as performed in particular patients was weak (r=0.27; p<0.05). Following cases of test results were found: 1. both results normal (1st NT, 2nd NT) in 117 patients (15.0 %), 2. 1st result normal, while the 2nd one falsely decreased (1st NT, 2nd FD) in 283 (36.3 %), 3. reverse situation (1st FD, 2nd NT) in 62 (7.9 %), 4. both results decreased in 318 (40.8 %), either truly (1st TD, 2nd TD) or falsely (1st FD, 2nd FD) - with unknown incidence. The following incidences (probabilities - P) result from the above data: A. P(1st NT, 2nd NT) = P(1st NT) intersection P(2nd NT) = 0.150; B. P(1st NT, 2nd FD) = P(1st NT) intersection P(2nd FD) = 0.363; C. P(1st FD, 2nd NT) = P(1st FD) intersection P(2nd NT) = 0.079. The expression (B intersection C)/A was calculated by transforming the right side of the equations: (B x C)/A = 0.192. Thus, 19.2 %, out of all 780 patients, i.e., 150 children (47.2 % of 318 patients, assessed as GH-deficient), could have both test results FD. CONCLUSIONS The observed differences in GH response to particular pharmacological stimuli, as well as the high incidence of falsely decreased GH response in the two stimulating tests, should be taken into account in qualifying short children to GH therapy.