Purpose The purpose of this paper is to identify locally relevant factors that predict low haemoglobin (Hb < 11.0g/dl) level of pre-school children (6–23 months) in an impoverished area of Ghana. Design/methodology/approach This study used a cross-sectional analysis of 278 mother–child pairs selected from households using random sampling techniques. Child feeding knowledge, antenatal care (ANC) practices and socio-demographic characteristics were assessed. Hb levels were laboratory determined. This paper modelled the predictors of low Hb in multiple linear regressions. Findings More than half (53.0%) of the children were female and aged more than one year (60.4%). Mean Hb was 8.5 ± 1.5 g/dl, and the prevalence of anaemia was 95.0%. High child feeding knowledge was independently associated with increased Hb of children after adjusting for the effects of other factors (β = 0.61, p < 0.005), and mothers who made at least four ANC visits during pregnancy had children with high Hb (β = 0.42, p = 0.04). This paper found some evidence of an inverse association between high household wealth and child Hb levels (β= −0.75, p = 0.06). Originality/value This study reports a high prevalence of low haemoglobin levels and anaemia in pre-school children in an impoverished area in Ghana. Maternal education on appropriate child feeding and encouragement of antenatal care attendance during pregnancy are key factors to increasing child haemoglobin levels in this setting. Though the current study bridges the knowledge gap between health service education and practice, a large population-based study is required to confirm the finding.