OBJECTIVE:: This study was carried out to evaluate the quality of the contents of referral letters received at the pediatric emergency unit of the University College Hospital, Ibadan, Nigeria. DESIGN:: We prospectively reviewed consecutive referral letters received over a 6-month period. The details of the contents of each letter were recorded using a structured proforma by 2 of the investigators after consent was obtained from the parent or caregiver. RESULTS:: There were 974 patients admitted with referral letters; this accounted for 54.8% of all admissions. There were 568 boys and 406 girls (ratio, 1.4:1). More than one tenth of the referred patients reported after 24 hours of writing the letters. Letters were written by physicians (69.2%), registered nurses (21.3%), hospital assistants (2.1%), traditional birth attendants (0.4%), and non-health workers (0.3%). The identity of the writers of 65 letters (6.7%) could not be defined. More than half of the letters did not contain the patients’ age, the treatment given, the findings from the investigations performed, the medical history, and what the writers expect from the referral. Other missing information includes examination findings (47.9%), provisional diagnosis (38.6%), history of presenting complaints (36.6%), writers’ addresses (32.5%), reasons for the referral (23.9%), patients’ sex (20.1%), and patients’ names (3.4%). The most frequently stated reason for referral was poor or no response to the treatment given (17.8%). CONCLUSION:: The contents of referral letters from the general practitioners to the pediatric emergency unit were found to be grossly inadequate. To enhance the quality of correspondence between the referring physicians and pediatricians, there may be a need for training and introduction of letter-format prompt forms.