BACKGROUND: Most children with Clostridium difficile infection (CDI) experience community onset of CDI symptoms. METHODS: We retrospectively compared hospital-onset healthcare facility-associated CDI cases to community-associated (CA) CDI cases diagnosed by Cepheid Xpert tcdB polymerase chain reaction (PCR) at an academic children's hospital over a 1-year period. Saved stools from CDI cases additionally underwent anaerobic stool culture and multiplex gastrointestinal pathogen PCR testing. RESULTS: Compared with 25 hospital-onset healthcare facility-associated CDI cases, the 74 CA-CDI cases were more frequently <2 years old (18% vs. 0%, P = 0.034) and less frequently had antibiotic exposure in the past 30 days (26% vs. 88%, P < 0.0001), proton pump inhibitor exposure (16% vs. 36%, P = 0.036) or a gastrostomy tube (11% vs. 32%, P = 0.013). Among children diagnosed with CA-CDI, 19 (26%) had no identified CDI risk factors (immunocompromised; gastrostomy tube; recent antibiotic, proton pump inhibitor or inpatient/outpatient healthcare exposures). Clinical testing for viral pathogens was uncommon among children thought to have CA-CDI. Multiplex PCR testing of saved stool samples failed to identify C. difficile among 23% of cases diagnosed with CA-CDI by the Cepheid Xpert tcdB PCR assay. CDI antibiotic therapy was provided to nearly all patients testing positive by tcdB PCR irrespective of CDI risk factors. CONCLUSIONS: Many children diagnosed with CA-CDI by PCR lack CDI risk factors and have discordant results when additional CDI testing methods are performed, suggesting overdiagnosis of CDI in children with community-onset diarrhea. More selective CDI testing of low-risk pediatric patients is needed to more accurately diagnose CDI and limit unnecessary CDI antibiotic treatment in children.