Social inequalities in a variety of indicators of child health were measured using a 'small area' geographical method of social classification. Cross sectional analyses of routine child health information and of a population survey of the height of primary school children were used. Social classification was by census enumeration district of residence using the Townsend deprivation score. Over 21,000 children resident in Northumberland born between January 1985 and September 1990, and 9930 children aged 5-8.6 years in Northumberland schools were studied. The following differences between the most deprived 10% of areas and the most affluent 10% of areas were used as outcome measures: the proportion of birth weights less than 2800 g; the proportion of births to teenage mothers; the proportion of 15 month old children not immunised against pertussis; the proportion of infants not screened at 6 weeks of age; the proportion of children not screened at 18 months of age; and the mean height of children in SD scores. Between the most deprived and most affluent areas birth weights less than 2800 g varied from 18 to 11%, the percentage of teenage mothers from 18 to 3%, non-immunised children from 30 to 19%, children not screened at 18 months from 21 to 14%, and mean height from -0.2 SD scores to +0.1 SD scores. The area variation in screening at 6 weeks of age was less, but still poorer in deprived areas. It is concluded that small area methods are effective in showing inequalities in child health, even in a rural area where such methods might be expected to perform less well. Social inequalities in all the aspects of child health measured remain evident.