Understanding habitat preferences for endangered species is a high priority for management strategies to ensure minimum conflict between human uses and wildlife conservation. The purpose of this study was to identify oceanographic variables that predict occurrences of humpback whales Megaptera novaeangliae within the Cordell Bank and Gulf of the Farallones National Marine Sanctuaries, California, USA, to assess potential conflict with vessel traffic. We used data collected by Applied California Current Ecosystem Studies (ACCESS) conducted from 2004 to 2011. Using zero-inflated negative binomial regression, we developed predictive models and identified locations highly used by whales to characterize humpback whale habitat. We analyzed whale encounter rates at 3-km bin intervals in relation to bathymetric, surface and midwater hydrographic predictor variables and temporal variables characterizing oceanographic conditions. Our models included variables that accounted for detectability of whales. Two models were compared and contrasted: (1) a surface-only model, using only surface oceanographic variables, and (2) a surface + mid-water model, using both surface and mid-water variables. The surface + mid-water model performed significantly better than the surface-only model, which underestimated the amount of suitable whale habitat in the northern half of our study area. We compared resulting predicted habitat areas with previous and current San Francisco Bay Area shipping lane poly gonal footprints to investigate whether newly accepted changes in routes reduced areal overlap with humpback whale habitat. Although our analyses show that the area occupied by shipping traffic has decreased in areas of high predicted humpback whale habitat use, changes in vessel lane footprints do not account for several important aspects of ship-strike risk, including vessel frequency, speed, size and density patterns within the shipping lanes and variability between lanes.