To determine whether systematic differences exist between hypertensive children referred for evaluation by primary care providers and children identified through community-based screening, cardiovascular risk factors and surrogate markers of hypertensive injury were compared based on subject source (referral versus screening). Children referred to a hypertension clinic for persistently elevated blood pressure were compared with children identified as hypertensive during school screening of 5102 students in Houston public schools. M-mode echocardiography of the left ventricle was performed and subsequently reviewed by 2 independent sonographers blinded to identifying subject information. Subsets of subjects also underwent carotid artery ultrasound for measurement of intimal-medial thickness, overnight urine collections for microalbuminuria, and fasting serum cholesterol, triglycerides, and glucose. Ninety-seven total subjects (54 screening and 43 referral) met inclusion criteria and had technically adequate echocardiography performed. The prevalence of left ventricular hypertrophy (LVH) was 37%. Referral subjects demonstrated significantly greater left ventricular mass index (38.8 versus 34.2 g/m(2.7); P<0.01) and a higher prevalence of LVH (49% versus 28%; P<0.05). Among subjects who underwent carotid ultrasound (n=75), carotid intimal-medial thickness was significantly higher in referral subjects (0.61 versus 0.57, P<0.05). When controlling for BMI z score, which was significantly higher in referral subjects, systematic differences by subject source did not persist. These findings suggest that hypertensive children who are predominantly overweight, independent of the manner in which patients come to medical attention, will manifest evidence of more severe cardiovascular disease assessed by surrogate markers such as left ventricular mass index or carotid artery intimal medial thickness.