This paper reports the results of a blinded study comparing videofluoroscopy with bedside clinical evaluations by speech/language pathologists in the diagnosis of aspiration. One hundred and seven inpatients from a general rehabilitation hospital were evaluated over a four-month period. Of the total patient population, 43 (40%) aspirated at least one consistency of food during videofluoroscopy. Bedside evaluation identified only 18 (42%) of these patients. The positive predictive value of bedside assessment was 0.75; negative predictive value was 0.70. Aspirators on videofluoroscopy were more likely to have brainstem or multilobe central nervous system involvement than nonaspirators. However, there was no statistically significant difference in lesion sites between clinically detected and "silent" aspirators. While the significance of aspiration noted on videofluoroscopy is debatable, it is clear that bedside evaluation alone underestimates the frequency of aspiration in patients with neurologic dysfunction.