Antecedent soil moisture determines how much storage remains in bioretention media before rainfall, but its relationship with runoff control is poorly quantified. Three wet seasons of monitoring were analyzed for two field-scale bioretention systems with contrasting media depths, runoff loading, and drainage boundaries. For each event, antecedent available storage was calculated from the initial soil moisture profile relative to field capacity and, separately, to the 95th percentile of peak water content at each depth. Event-activated storage was defined as the greatest synchronous increase in water stored across the monitored profile following rainfall onset. Both storage metrics were expressed as equivalent rainfall depths over the contributing roof area. Runoff volume reduction ranged from 62.3% to 100% in BR1 and from 43.3% to 100% in BR2. Median event-activated storage over the media area was 4.9 mm in BR1 and 19.1 mm in BR2. However, after conversion to the contributing roof area, median antecedent storage was larger in BR1 than in BR2 under both the field capacity reference (3.8 versus 0.6 mm) and the high water reference (11.0 versus 5.0 mm). Greater antecedent storage was associated with a lower probability of drainage activation in both systems and greater runoff volume reduction in BR1. In BR2, greater event-activated storage was associated with lower runoff volume reduction, indicating that stronger within-event media wetting did not necessarily correspond to greater facility-scale runoff retention. Dynamic storage should therefore be interpreted in relation to hydraulic loading, media configuration, and drainage pathways.