The goal of this study was to compare the diagnostic value of an absent gag reflex in acute stroke patients with the bedside swallowing assessment (BSA) and assess its relationship to outcomes. Two hundred forty-two acute stroke patients had their gag reflex tested and a BSA performed. Numbers needing nasogastric or gastrostomy tube insertion were noted, also their discharge destination, discharge Barthel Index, and mortality. The mean age of the subjects was 76.5+/-10.2 years; 37.6% were male; 41.7% of the patients were dysphagic on BSA; 18.2% had an absent gag. Dysphagia was present in 88.6% of the patients with an absent gag and in 31.3% of those with an intact gag. The gag reflex was absent in 38.6% of dysphagic and 3.5% of nondysphagic patients. Comparing an absent gag against the criterion of the BSA, its specificity was 0.96, sensitivity 0.39, positive predictive value 0.89, and negative predictive value 0.69. Regression analyses found that an intact gag gave an Odds Ratio [CI] of 0.23 [0.06-0.91] for gastrostomy feeding but did not predict other outcomes. We conclude that the gag reflex is as specific as but less sensitive than the BSA in detecting dysphagia in acute stroke patients. An intact gag may be protective against longer-term swallowing problems and the need for enteral feeding.