We describe a method to convert continuously collected time–depth data from archival time–depth recorders (TDRs) into activity budgets for a benthic-foraging marine mammal. We used data from 14 TDRs to estimate activity-specific time budgets in sea otters (Enhydra lutris) residing near Cross Sound, southeast Alaska, USA. From the TDRs we constructed a continuous record of behavior for each individual over 39–46 days during summer of 1999. Behaviors were classified as foraging (diving to the bottom), other diving (traveling, grooming, interacting), and nondiving (assumed resting). The overall average activity budget (proportion of 24-hr/d) was 0.37 foraging (8.9 hr/d), 0.11 in other diving (2.6 hr/d), and 0.52 nondiving time (12.5 hr/d). We detected significant differences in activity budgets among individuals and between groups within our sample. Historically, the sea otter population in our study area had been expanding and sequentially reoccupying vacant habitat since their reintroduction to the area in the 1960s, and our study animals resided in 2 adjacent yet distinct locations. Males (n = 5) and individuals residing in recently occupied habitat (n = 4) spent 0.28–0.30 of their time foraging (6.7–7.2 hr/d), 0.17–0.18 of their time in other diving behaviors (4.1–4.3 hr/d), and 0.53–0.54 of their time resting (12.7–13.0 hr/d). In contrast, females (n = 9) and individuals residing in longer occupied habitat (n = 10) spent 0.40 of their time foraging (9.6 hr/d), 0.08–0.09 of their time in other diving behaviors (1.9–2.2 hr/d), and 0.51–0.52 of their time resting (12.2–12.5 hr/d). Consistent with these differences, sea otters residing in more recently occupied habitat captured more and larger clams (Saxidomus spp., Protothaca spp., Macoma spp., Mya spp.,Clinocardium spp.) and other prey, and intertidal clams were more abundant and larger in this area. We found that TDRs provided data useful for measuring activity time budgets and behavior patterns in a diving mammal over long and continuous time periods. Fortuitous contrasts in time budgets between areas where our study animals resided suggest that activity time budgets estimated from TDRs may be a sensitive indicator of population status, particularly in relation to prey availability.