Monitoring variation in populations of coastal dolphins presents a range of challenges. Many species occur at low local population levels, are cryptic and appear to range over larger areas than presumed. Here we present the results of a three and half year intensive monitoring study undertaken in Darwin Harbor and two neighboring sites (1086 km2). The study comprised multi-site robust design capture-recapture models that yielded estimates of abundance, apparent survival and temporary emigration on three species of coastal dolphins: Australian humpback (Sousa sahulensis), bottlenose (Tursiops sp.) and Australian snubfin (Orcaella heinsohni). Combining all three sites, abundance estimates varied between species. The Australian humpback was the most abundant with a mean of 90, bottlenose were stable at mean of 27 and the Australian snubfin varied widely from 19 to 70 with a mean of 41. Overtime, Australian humpback abundance estimates showed a steady decline in Darwin Harbor but a population increase in the two neighboring sites was recorded, suggesting there were movements out of Darwin Harbor. However, the estimates of movement rates were not sufficiently sensitive to demonstrate this, due to the relatively small size of the local population and consequent low rates of observed movement. The multi-state robust design model offers the potential for assessing abundance estimates and population trends. It is able to distinguish between movements to and from a site from demographic changes on the site that otherwise might be attributed to other factors (i.e., decrease in survival). The study highlights the substantial effort and time required to detect population trends for coastal dolphins by needing to account for movement among sites. However, the reality of assessing the conservation status for coastal dolphins is challenging. Moreover, to enact conservation measures a reassessment at both global and nationals levels of the IUCN Red List categories (A and C) is required.