Background Pneumonia accounts for 16% of all child deaths in Kenya, and it is the second killer of children under the age of five. Pneumonia burden is attributed to inadequate access to healthcare and poor health-seeking by caregivers. Endebess caregivers report delays of 1-14 days and subject their children to medicine from shops before visiting a hospital, as evidenced by the history of the presenting illness in patient files. In addition, irrational prescription has been demonstrated regardless of clinical guidelines. This study explored health facility practices and drug prescription patterns that influence health-seeking behaviour in Endebess Sub-county, Kenya. Methods A cross-sectional study involving 273 caregivers accompanied by a qualitative study on 24 health personnel was conducted from May to August 2019. Using a simple random sampling technique, a questionnaire was administered to caregivers visiting the health facilities. Key informant interview guides were also conducted with health care personnel. Quantitative data were analysed using logistic regression, whereas qualitative data were analysed by searching for emerging themes on health facility practices and prescription patterns across health facilities. Results Overall, stratification by health facility practices revealed that there were numerous significant results at the level of P<0.001, including the availability of drugs, administration of a starting dose, type of drug prescribed, willingness to recommend the facility, place caregiver seeks treatment (P=0.005) and facility of choice (P=0.001), which all influenced willingness to revisit the hospital. In relation to health information, the significant results at the level of P<0.001 included adequate instructions, the confidence of the caregiver to give the correct dosage at home, satisfaction levels and shared decision making on choices of antibiotic therapy (P=0.005) were associated with a higher willingness to revisit the hospital. There was a varied pattern of prescriptions for antibiotics. Consistently the odds of willingness to revisit the hospital increased among caregivers who had an injection medication prescribed for their children (P=0.380; odds ratio, OR=2.06, 95% confidence interval, CI=0.41-10.42) relative to those on oral medication. Conclusions There is a need for strict control of prescribing antibiotics in children under five years. Shared decision making on the choice of therapy should not be limited to chronic illnesses.