The 2,680 km of maritime coasts contained within Glacier Bay National Park and Preserve (GLBA), Katmai National Park and Preserve (KATM), and Wrangell-St. Elias National Park and Preserve (WRST) in Alaska represent some of the most remote, pristine coastal areas in the United States. The apex biological resources such as marine mammals, seabirds, salmon, and bears that are so much of the Park experience depend on healthy prey bases and habitats. Intertidal and subtidal invertebrates and plants provide not only structure in these coastal communities, but many of the invertebrates are important components of the diets of these higher-order consumers. However, these biologically rich and productive communities are facing increased anthropogenic pressures. At the present time, it is difficult to determine if observed changes in species or variations in habitat are a result of natural forces or anthropogenic factors. The ability to detect change is related to the extent of variation that exists. Variation may occur in all aspects of populations and communities (e.g., species distribution, size and age structure, abundance, birth rates, mortality rates, dispersal abilities, composition of communities, diversity, strengths of interactions within communities). As variation in any parameter increases, the ability to detect change in that parameter (and the population or community in question) decreases, assuming equal sampling effort. Variation may be reduced and the power to detect change increased by stratifying the populations and communities along known gradients of variation, or by increasing sampling effort. Increased sampling costs more, while stratified sampling reduces the ability to make inferences beyond the stratum sampled. Managers will need to make a series of assessments and decisions prior to designing a monitoring program. These include: what level of inventory and knowledge of variation in the coastal communities is known; the level of change that they want to be able to detect; whether they want to focus on particular habitats, species or communities; whether they want to take a population approach; what risks they are most concerned about in the near-term; what long-term risks they want to take into account; which species, assemblages, or communities would be sensitive ecological indicators of change; do they want results of monitoring to be able to be generalized to zones, habitats, or the entire park coastline; how do they expect to make the program sustainable; what resources, both human and financial, are available for long-term monitoring; and how frequently do they want to monitor resources. We propose to provide managers with a cost-benefit analysis of several different monitoring designs (broad-scale inferential; nested inferential [selected habitat, coarse-grained]; and intensive sampling [selected habitat, fine-grained]). It is possible to incorporate at least several levels into a nested design with the purpose of quantifying both the costs associated with the design elements and the power to detect change. This analysis, combined with further discussion, should facilitate the decisions that managers need to make in designing a long-term monitoring program.