We report on 49 boys with constitutional growth delay (CGD) who were initially seen in our clinic at a mean chronological age of 13.3 years (range, 7.3-16.4) and a bone age of 11.1 years (range, 6.0-13.5). All were below the 5th height percentile for chronological age. A positive family history with delayed growth and puberty in one or both parents could be elicited in 75%. All 49 patients were re-examined at a mean age of 22.9 years (range, 20.4-31.2). Measured final height was 171.3 cm (range, 161.2-181.7), which was slightly, but significantly lower than mean target height of 173.0 cm. Final height expressed as standard deviation score (SDS) of a male adult population standard was -1.0 (range, -2.4 to 5), also significantly lower than initial height SDS related to bone age (SDSBA) of -0.5 (range, -1.6 to 2). If related to target height (Tanner), final height was found to correlate positively with the initial bone age deficit and the initial height SDSBA. Observed final height was also compared with the predicted adult height by the methods of Bayley-Pinneau (BP), Tanner-Whitehouse Mark II (TW II) and Roche-Wainer-Thissen. Regression equations between all three prediction methods and final height showed an excellent correlation (P < 0.0001). However, only by the BP method was predicted height very close to and no different from measured final height (paired t-test). Despite this, final height in 16 of 49 patients (32.6%) differed by more than 5.0 cm from BP predicted height.(ABSTRACT TRUNCATED AT 250 WORDS)