BACKGROUND & AIMS: Measurements of hepatitis B surface antigen (HBsAg) and hepatitis B virus (HBV) DNA might help to identify carriers of inactive HBV. We assessed the performance of repeated measurements of HBsAg over a median time period of 8 years. METHODS: We performed a retrospective study of 292 HBe antigen-negative patients with chronic HBV infection, normal levels of alanine aminotransferase (ALT), levels of HBV DNA <20,000 IU/mL, and no cirrhosis who visited the outpatient clinics at 8 tertiary care centers in Europe, Asia, and Australia from 1990 through 2011. Patients were determined to be carriers of inactive HBV (level of HBV DNA <2000 IU/mL and serum levels of ALT that remained normal) or to have HBV activity (level of HBV DNA fluctuating >2000 IU/mL and/or abnormal levels of ALT) after each year of follow-up. Patients were followed for a median time of 8 years (range, 4-9 years). Dynamic regression analysis was used to study changes in level of HBsAg and HBV phase and to update the risk of HBV activity. RESULTS: One year after study enrollment, 189 patients (65%) had inactive HBV and 103 patients (35%) had HBV activity. Based on dynamic analysis, the probability that a patient would have HBV at any following year differed according to level of HBsAg; odds were 97% for patients with initial level of HBsAg <100 IU/mL, 85% for patients with initial levels 100-1000 IU/mL, and 76% for patients with initial levels >1000 IU/mL (P < .001). Having inactive virus for any 2 consecutive years predicted having inactive virus in any third year. However, 15% of patients with level of HBsAg >100 IU/mL had HBV activity in the third year. The combination of HBsAg level <100 IU/mL and HBV DNA level <2000 IU/mL identified patients whose virus remained inactive for the entire follow-up period, with 98% specificity and a positive predictive value of 97%, for all HBV genotypes. Patients with HBV activity who had levels of HBV DNA < 5000 IU/mL and decreases in HBsAg of 0.5 log IU/mL or more for 1 year had a high probability of becoming carriers of inactive HBV in the next year. CONCLUSIONS: In a retrospective, dynamic analysis of almost 300 patients with chronic HBV infection, we found that levels of HBsAg <100 IU/mL identify patients with inactive virus with a high level of specificity. HBsAg levels should therefore be used to define phases of HBV infection in HBe antigen-negative patients.