This prospective cohort study compared measurements of maternal home blood pressure (HBP) with clinic blood pressure (CBP) before 20 weeks' gestation to determine associations with the risk of delivering a lower birth weight infant. A total of 605 Japanese women were included. Exposures were initial CBP, made between 10 weeks 0 days and 19 weeks 0 days, and HBP for comparison made within 1 week of CBP. Outcome was infant's birth weight, categorized and ranked as follows: ⩾3500 g, 3000-3499 g, 2500-2999 g and <2500 g. The proportional odds model with possible confounding factors was applied to compare the associations between CBP and HBP on infant birth weight. When both CBP and HBP were included simultaneously, the adjusted odds ratios (ORs) per 1 standard deviation (1s.d.) increase in clinic and home diastolic BP (DBP) were 1.06 (95% confidence interval (CI): 0.87-1.30) and 1.28 (95% CI: 1.04-1.58), respectively. The adjusted ORs per 1s.d. increase in clinic and home mean arterial pressure (MAP) were 1.02 (95% CI: 0.83-1.24) and 1.29 (95% CI: 1.04-1.59), respectively. Systolic BP measurement was not associated with infant birth weight. In conclusion, high maternal home DBP and MAP before 20 weeks' gestation was associated with a higher risk of lower infant birth weight than clinic DBP and MAP. Therefore, in addition to CBP, it may be worth having pregnant women measure HBP to determine the risk of lower infant birth weight.