OBJECTIVES:: The burden of acute gastroenteritis (AGE) in US children is substantial. Research into outpatient treatment strategies has been hampered by the lack of easily used and validated gastroenteritis severity scales relevant to the populations studied. We sought to evaluate, in a US cohort, the reliability, construct validity, and generalizability of a gastroenteritis severity scale previously derived in a Canadian population, the modified Vesikari score (MVS). METHODS:: We conducted a prospective, cohort, clinical observational study of children 3 to 48 months of age with acute gastroenteritis presenting to 5 US emergency departments. A baseline MVS score was determined in the emergency department, and telephone follow-up 14 days after presentation was used to assign the follow-up MVS. We determined reliability using inter-item correlations; construct validity via principal component factor analysis; cross-sectional construct validity via correlations with the presence of dehydration, hospitalization, and day care and parental work absenteeism; and generalizability via score distribution among sites. RESULTS:: Two hundred eighteen of 274 patients (80%) were successfully contacted for follow-up. Cronbach alpha was 0.63, indicating expectedly low internal reliability because of the multidimensional properties of the MVS. Factor analysis supported the appropriateness of retaining all variables in the score. Disease severity correlated with dehydration (P < 0.001), hospitalization (P < 0.001), and subsequent day care (P = 0.01) and work (P < 0.001) absenteeism. The MVS was normally distributed, and scores did not differ among sites. CONCLUSIONS:: The MVS effectively measures global severity of disease and performs similarly in varying populations within the US health care system. Its characteristics support its use in multisite outpatient clinical trials.