Long-term research programs can benefit from early and periodic evaluation of their ability to meet stated objectives. In particular, consideration of the spatial allocation of effort is key. We sampled 4 species of stream salamanders intensively for 2 years (2010–2011) in the Chesapeake and Ohio Canal National Historical Park, Maryland, USA to evaluate alternative distributions of sampling locations within stream networks, and then evaluated via simulation the ability of multiple survey designs to detect declines in occupancy and to estimate dynamic parameters (colonization, extinction) over 5 years for 2 species. We expected that fine-scale microhabitat variables (e.g., cobble, detritus) would be the strongest determinants of occupancy for each of the 4 species; however, we found greater support for all species for models including variables describing position within the stream network, stream size, or stream microhabitat. A monitoring design focused on headwater sections had greater power to detect changes in occupancy and the dynamic parameters in each of 3 scenarios for the dusky salamander (Desmognathus fuscus) and red salamander (Pseudotriton ruber). Results for transect length were more variable, but across all species and scenarios, 25-m transects are most suitable as a balance between maximizing detection probability and describing colonization and extinction. These results inform sampling design and provide a general framework for setting appropriate goals, effort, and duration in the initial planning stages of research programs on stream salamanders in the eastern United States.