Growth hormone excretion can easily be measured in the urine using ultrasensitive methods. The large day-to-day variation has, however, restricted its diagnostic usefulness. The present study aimed to evaluate the individual variation of GH in the urine (uGH) during normal prepubertal growth. Eighty-four prepubertal normally growing children were followed monthly for 13 months. During this period, 3,207 overnight urine samples were collected. The urine collection time was unrelated to the uGH concentration (p > 0.05), while there was a significant negative correlation between the uGH concentration and urine volume (the Spearman correlation coefficient of –0.33, p < 0.0001), while the calculated excretion of GH in the urine showed a positive correlation with the urine volume (r = 0.35; p < 0.0001). A reference chart, based on SD scores, was developed in order to avoid this volume dependency and to optimally normalize the skewed distribution of the uGH concentrations. The use of this model reduced the individual day-to-day variation of uGH from a coefficient of variation of 43 to 21%. Differences in mean cross-sectional urinary GH concentration was found between different months exceeding the expected methodological variation. This variation showed no seasonal pattern. Only 0.2% of triplicate values (three consecutive overnight uGH values) were all below –2 SD scores and 0.1% were above +2 SD scores. The mean uGH SD score for the boys was 0.01 (SD = 0.98), which was similar to that for the girls (–0.04; SD = 1.06). We found that uGH excretion can be estimated in a more robust way, using a SD score based reference chart that handles both the positive correlation between urinary GH and urine volume and the skewed distribution of urinary GH. This model reduced the day-to-day variability of uGH by half. Overestimation of GH in large urine volumes may be due to increased gradient between GH in urine and serum following increased urine volumes.