Eighty per cent of all identified children 2 1/2-5 1/2 years old in a total geographic community in Bogalusa, Louisiana, were examined for cardiovascular disease risk factor variables, including blood pressure (BP), anthropometric measurements, and blood lipids. Blood pressure was measured by three instrument types, each with three readings, according to a rigid, randomized design. Results from the three instruments were compared, and potential biases for each instrument are listed. All observations on the children were analyzed by multiple regression with BP as the dependent variable. After controlling for the mood of the child, the authors found that some index of body size was positively related to BP, whereas age, race, and serum lipids were not consistently related. The regressions of blood pressure on height as well as on log weight fit straight lines. These results conform with the findings for the 5- to 14-year-olds in this community. The basal or fundamental reference blood pressure is likely to rise gradually from birth to the age of incipient adult stature, and to be linearly related to height and to log weight. Changes observed over time are needed to relate BP levels in childhood to the early natural history of essential hypertension.