Hepatitis B is a major global public health threat. According to China's National Notifiable Disease Reporting System (NNDRS), Gannan Tibetan Autonomous Prefecture (Gannan) had the highest incidence of hepatitis B in Gansu Province during 2004 to 2016. We evaluated NNDRS hepatitis B case reports from Gannan to determine accuracy of diagnosis and to understand factors associated with inaccuracy. We reviewed medical records with hepatitis B diagnosis hospitalized in seven county hospitals in Gannan between January 1, 2016 and July 31, 2017. Using national "Classification and Diagnostic Procedures for Hepatitis B," we independently reclassified the diagnoses. We determined the positive predictive value (PPV) of reported hepatitis B cases. We investigate clinicians' understanding of the diagnostic and reporting criteria for hepatitis B by questionnaire. We reviewed and re-categorized 400 inpatients reported. Sixteen cases had been reported as acute hepatitis B, but on re-categorization, none were acute hepatitis B cases. PPVs for chronic hepatitis B and unclassified hepatitis B cases were 66% and 15% respectively; 327 (82%) of the reported hepatitis B cases were inaccurately classified; 261 were carriers, 59 were reported previously, and 7 did not have hepatitis B. The actual incidence of hepatitis B in Gannan in 2016 was estimated to be 19/100,000, significantly below the reported incidence of 106/100,000. Among reported cases, 81% had been tested for Alanine aminotransferase, 52% for hepatitis C antibody, 80% with liver ultrasound, 32% for hepatitis A antibody, and 7% for hepatitis B virus (HBV) DNA. Not all cases were tested for anti-HBc IgM or hepatitis E antibody or had a liver biopsy. In the knowledge test, 56% of clinicians accurately diagnosed three simulated cases of acute hepatitis B, and 17% correctly diagnosed two simulated cases chronic hepatitis B; 22% knew that "a client with only HBsAg positivity need not be reported." The falsely high incidence in Gannan was due to diagnostic and reporting inaccuracies. We recommend that clinicians and laboratorians receive additional training in hepatitis B diagnostic criteria and reporting standards, including appropriate use of IgM anti-HBc tests. Hepatitis B surveillance data should be periodically reviewed and evaluated for accuracy.