Study objective: To determine practice patterns regarding administration of the "GI cocktail" (a mixture of liquid antacid, viscous lidocaine, and an anticholinergic) in the emergency department in a single hospital and the responses and final dispositions of patients who received the cocktails. Design: A retrospective review of ED charts. Setting: Urban university hospital ED with an annual census of 50,000 visits. Participants: Ninety-seven consecutive patients who received a GI cocktail in the ED. Results: Forty-nine patients (50%) received a GI cocktail for a chief complaint of abdominal pain and 40 (41%) for a chief complaint of chest pain. The reason for administration of a GI cocktail was documented on only one chart. Sixty-six patients (68%) received at least one other drug, at a median time of 9 minutes before administration of the GI cocktail. The most commonly coadministered drug was a narcotic (56 patients), followed by nitroglycerin (22 patients), antiemetics (13 patients), H₂-blockers (13 patients), and aspirin (10 patients). Thirty-three patients (34%) had symptomatic relief with the cocktail alone, 35 (36%) had symptomatic relief with the cocktail plus other drugs, 7 (7%) had no response to the GI cocktail alone, and 5 (5%) had no response to the cocktail with other drugs. In 17 patients (18%) the response was not documented. Chest pain patients and abdominal pain patients had a similar frequency of response. There was also similarity of response between admitted and discharged patients. Conclusion: Although symptomatic relief after administration of a GI cocktail is often noted, it is difficult to differentiate the effects of the cocktail from those of other coadministered medications such as morphine or nitroglycerin. We urge ED physicians to use the GI cocktail in a rational manner. [Wrenn K, Slovis CM, Gongaware J: Using the "GI cocktail": A descriptive study. Ann Emerg Med December 1995;26:687-690.]