Military landscapes represent a mixture of undisturbed natural ecosystems, developed areas, and lands that support different types and intensities of military training. Research to understand water-quality influences of military landscapes usually involves intensive sampling in a few watersheds. In this study, we developed a survey design of accessible headwater watersheds intended to improve our ability to distinguish land-water relationships in general, and training influences, in particular, on Fort Stewart, GA. We sampled and analyzed water from watershed outlets. We successfully developed correlative models for total suspended solids (TSS), total nitrogen (TN), organic carbon (OC), and organic nitrogen (ON), which dominated in this blackwater ecosystem. TSS tended to be greater in samples after rainfall and during the growing season, and models that included %Wetland suggested a "build-and-flush" relationship. We also detected a positive association between TSS and tank-training, which suggests a need to intercept sediment-laden runoff from training areas. Models for OC showed a negative association with %Grassland. TN and ON both showed negative associations with %Grassland, %Wetland, and %Forest. Unexpected positive associations were observed between OC and equipment-training activity and between ON and %Bare ground + Roads. Future studies that combine our survey-based approach with more intensive monitoring of the timing and intensity of training would be needed to better understand the mechanisms for these empirical relationships involving military training. Looking beyond local effects on Fort Stewart streams, we explore questions about how exports of OC and nitrogen from coastal military installations ultimately influence estuaries downstream.