We evaluated the growth hormone (GH) response to an acute clonidine test (0.15 mg/m2 po) in 30 normal prepubertal children (stature between the 3rd and 97th centile), in 29 short children (stature < 3rd centile for age) with height velocity (HV) > 10th centile and in 20 short children with HV < 10th centile. The three groups had comparable chronological ages. After clonidine administration mean peak GH levels were similar in the three groups (19.4 +/- 9.8, 17.7 +/- 8.8 and 14.6 +/- 8.9 micrograms/l, mean +/- SD, respectively). By choosing 10 micrograms/l as the limit for a normal response we found that stimulated GH levels had a sensitivity of 50% and a specificity of 83% in identifying children with suspected GHD (short children with subnormal HV). The diagnostic accuracy was almost superimposable, for cut-off values of 10 and 12 micrograms/l. Eight of the 10 children with subnormal HV and a GH peak < 10 micrograms/l had a GH peak < 10 micrograms/l also after a second stimulation test. Six of the 29 short children with normal HV had a GH peak < 10 micrograms/l. Only one of them had a GH peak < 10 micrograms/l after a second stimulation test. Five of the normal children had peak GH levels < 10 micrograms/l. These results indicate that HV is a useful variable to predict the GH response to an acute GH stimulus, since the great majority of children with a normal growth rate had a normal GH response to at least one stimulation test.