Earlier identification of potentially efficacious treatments in early development trials requires on-treatment response markers. We hypothesized that mean week 12 or 24 HBsAg decline could be a useful marker for subsequent off-treatment sustained HBsAg clearance at the treatment arm level in HBV trials. We used individual patient data from the studies HBV 9901 (peginterferon [PEG-IFN] versus PEG-IFN+lamivudine for HBeAg-positive CHB), PARC (PEG-IFN±ribavirin for HBeAg-negative CHB) and published data from 0149 (PEG-IFN±tenofovir for HBeAg-positive and HBeAg-negative CHB) and LIRA-B (PEG-IFN for HBeAg-positive CHB) to define the relationship between mean week HBsAg decline and HBsAg loss at 6 months post-treatment. A within-study comparison of HBsAg decline at weeks 12 and 24 between patients with or without HBsAg clearance was used to make projections beyond the observed HBsAg data. Across trials, a more pronounced mean HBsAg decline at week 24 was associated with higher rates of subsequent HBsAg loss. Mean HBsAg decline data at week 24 for patients with or without HBsAg clearance from HBV 9901 (4.3 vs 0.5), PARC (4.8 vs 0.3) and 0149 (PEG-IFN+TDF arm; 4.6 vs 0.6) were used to extrapolate this relationship beyond observed rates of HBsAg. An additional mean 1 log10 decline at week 24 versus a comparator arm is expected to translate into a 20%–30% increase in subsequent HBsAg loss during off-treatment follow-up. Observations were similar for week 12 data, but the relationship was less strong. Mean week 24 HBsAg decline predicts subsequent HBsAg loss and could be a valuable and useful early endpoint in HBV-treatment trials.