Fifteen girls with Turner syndrome (TS) were submitted to GH secretion assessment before undergoing hGH therapy. In the first 9 months, hGH was given at a dose of 0.5 IU/kg/week s.c. daily; afterward, the dose was increased to 1 IU/kg/week s.c. daily. The girls were prepubertal, with a mean (SD) chronological age (CA) of 12.5 (2.6) years, and a mean (SD) bone age of 10.5 (1.8) years. A clonidine stimulation test, 1-29 GHRH test and GH spontaneous nocturnal secretion assessment were performed in all patients. Results showed a variable pattern of GH secretion in 10 patients, in only 2 did we find all values definitely normal, and in 3 we found a total GH deficiency. Height velocity, expressed as standard deviation scores (SDS) for CA according to Turner references, during the first year of treatment increased significantly: 0.36 (1.15) -3.30 (2.87) (p < 0.001), and the increment remained quite unchanged during both the second and third years: 3.16 (2.96) and 2.55 (3.87), respectively (n.s.). Height, expressed in SDS for CA for Turner references, increased significantly throughout the whole period of treatment and reached the highest value at the end of the third year of therapy. GH secretion parameters poorly correlated with pretreatment auxological data or response to treatment. Our long-term study confirms that in TS GH measurement is not useful in indicating hGH therapy or in predicting the response.