This study was carried out to examine the biological and environmental variables associated with non-organic short stature. We observed an unselected population of very short normal children (SN) and their age- and sex-matched controls (C) within the community. All 14,346 children in two health districts entering school during 2 consecutive years were screened for short stature, and those whose height lay below the 3rd centile, according to Tanner and Whitehouse standards (n = 180) were identified. Excluding 32 with pathology, five from ethnic minorities and three who refused to take part, the remaining SN children (mean height SDS-2.26) were matched with 140 age- and sex-matched controls (C) of average height (mean height SDs 0.14). Birth weight, target height and predicted adult height (based on parental height and bone age respectively), medical and social background (obtained from parental interviews), and school performance (assessed by class teachers) were the main outcome measures. Mean birth weight of the SN children was significantly lower than C (SN = 2845 g, C = 3337 g, P < 0.001). Mean mid-parental target height was also very different (SN = 162.0 cm, C = 170.9 cm, P < 0.001). Thirty-five per cent of SN children (C = 6%) had height SD scores below parental target range, though only 10% had predicted heights below target range (mean delay in bone age 0.68 years). There was a significant difference between SN children and C in the number of children in the household (SN = 2.8, C = 2.4 (P = 0.007) and in socio-economic status (P < 0.002). Many more SN children were in social classes IV and V (SN = 31%, C = 13%, P < 0.002), and had an unemployed father (SN = 22%, C = 10%, P < 0.010), highlighting the importance of environmental influences on growth. One in four SN children was judged to have serious psychosocial problems. However, the lower the socio-economic class, the less likely the SN children were to be inappropriately short for parents. Significantly more SN children were reported to have asthma (SN = 18%, C = 7%, P < 0.007) and eczema (SN = 19%, C = 5%, P < 0.001), though only the latter was significantly associated with stature below target height for both SN and C groups. Biological variables are often insufficient to explain short stature. No child, whatever the parental height, should be dismissed as normal without careful evaluation, as poor growth in the early years may be an important pointer to an adverse but potentially remediable environment.