Regional differences in adult morbidity and mortality within England (i.e., north-south divide or gradient) and between England and Scotland (i.e., Scottish effect) are only partly explained by adult levels of socioeconomic status or risk factors. This suggests variation in early life, and is supported by the foetal origins and life-course literature which posits that birth outcomes and subsequent, cumulative exposures influence adult health. However, no studies have examined the north-south gradient or Scottish effect in health in the earliest years of life. The aims of the study were: i) to examine health indicators in English and Scottish children at birth and age three to establish whether regional differences exist; and ii) to establish whether observed changes in child health at age three were attributable to birth and/or early life environmental exposures. Respondents included 10,639 biological Caucasian mothers of singleton children recruited to the Millennium Cohort Study (MCS) in the year 2000. Outcome variables were: gestational age and birth weight, and height, body mass index (BMI), and externalising behavioural problems at age three. Region/country was categorised as: South (reference), Midlands, North (England), and Scotland. Respondents provided information on child, maternal, household, and socioeconomic characteristics. Results indicated no significant regional variations for gestational age or birth weight. At age three there was a north-south gradient for externalising behaviour and a north-south divide in BMI which attenuated on adjustment. However, a north-south divide in height was not fully explained by adjustment. There was also evidence of a ‘Midlands effect’, with increased likelihood of shorter stature and behaviour problems. Results showed a Scottish effect for height and BMI in the unadjusted models, and height in the adjusted model, but a decreased likelihood of behaviour problems. Findings indicated no regional differences in health at birth, but some regional variation at age three supports the cumulative life-course model.