Objectives To obtain a general overview of infant and young child feeding practices in one rural county in China and identify current delivery channels and challenges.

Design A cross-sectional study.

Setting A rural county, Zhao County, in Hebei Province, China.

Participants 10 clusters were first selected within each township (16 townships in total) with proportional to population size sampling. In each cluster, a name list was used to select 13 children aged 0–23 months. We interviewed caregivers of all the selected children.

Primary and secondary outcomes measures Coverage of infant feeding practices, reasons for low coverage of infant feeding practices and current delivery channels of infant feeding practices.

Results Findings from our survey indicated that infant feeding practices were poor. Early initiation of breastfeeding was only 22.4%, exclusive breastfeeding for 6 months was less than 10% and continued breastfeeding up to the age of two was just 38.2%. Only 32.5% of children were given iron-rich or iron-fortified foods. The leading sources of infant feeding information were family members, neighbours, friends and popular media. Only around 20% of the information came from health facilities and nearly none came from communities. Household property data showed that 99.9% of households owned televisions and 99.4% owned mobile phones. In addition, 61.2% of the households owned computers, with 54.8% having access to the internet.

Conclusions Few caregivers of children in Zhao County received feeding information during pregnancy and after delivery. Moreover, their feeding knowledge and practices were poor. Multi-channel approaches, delivered through health facilities, community resources, popular media, the internet and mobile phones, hold potential to improve infant feeding practices and should be explored in future studies.

Strengths and limitations Although this study took place only within one county, a full range of globally standard feeding indicators was used to assess the feeding practices of caregivers in our study. The name lists of children in some villages may not be complete, and therefore selection bias may have occurred. Some indicators may have recall bias due to long recall time.

Trial registration number ChiCTR-PRC-11001446.