INTRODUCTION: In the referral process, referral papers are the standard and typically the sole method of communicating information between general practitioners and hospital specialists. Sub-optimal referral letter can be a source of poor continuity of care (delayed diagnosis, multiple medication, multi-drug resistance, high litigation risk, unnecessary testing and extra-medical costs) and therefore, decrease the quality of care. Referral papers of high quality are an essential part of good clinical care and act as the interface between health care professionals in primary, secondary and tertiary care. OBJECTIVE: The aim of this study is to assess the quality of documentation on referral papers of patients referred to Tikur Anbessa Specialized Hospital Emergency adult, pediatrics and gynecology and obstetrics departments. METHODOLOGY: This study was conducted at TASH EDs from December- June 2014 by implementing a retrospective cross sectional study design. A total of 1011 patient referral papers were recruited by simple random sampling method. Data was collected from patients’ individual folders retrospectively. For collecting relevant information, data was collected by using dummy tables and analyzed using SPSS version 20.0. RESULTS: All 1011 eligible referral letters from Tikur Anbessa Specialized Hospital Emergency Department were systematically assessed in this study. The result shows that the name of the patient featured in all of referral letter (100%, n=1011). Only 29.8% of referral letters bearing the patient’s address while 70.3% of referral letters contain history of present illness; 30.3% of referral letter contain physical examination and 19.4% of referral letters contain all the vital signs. The histories of allergies were reflected in none of the referral letters. About 12.2% referral letters were not entirely legible. CONCLUSION: Most of the socio-demographic data except the address were documented in the referral papers. The clinical information section (the most important part) of the referral paper was strikingly deficient especially history of allergy, vital signs, physical examination findings, chief complaint(s), results of basic investigations, treatment given. Only the working diagnosis and reason for referral were documented in most referral papers. In a quarter of referral papers assessed, the receiving unit was not mentioned, of which more than half wrote to any hospital. Signature of the referring clinician rather than name or qualification was documented