Stream-riparian networks are subject to multiple anthropogenic pressures that cause habitat and diversity losses, threaten ecosystem services, and drive stakeholder conflicts. The rehabilitation of woody riparian buffers along stream channels is frequently proposed as a management measure capable of addressing multiple goals, including enhancement of ecosystem functioning, and protection of biodiversity. However, the planning of riparian rehabilitation to simultaneously meet multiple goals requires a clear understanding of potential trade-offs, and a landscape-scale perspective on planning of reforestation measures. The BiodivERsA funded project CROSSLINK used data collected from spatially-explicit field sampling campaigns to develop a landscape optimization algorithm for allocating riparian forest management measures in the heavily agricultural catchment of the Zwalm River (Belgium). We optimized forest allocation to improve three target indicators, viz. stream ecological quality (EPT index), functional diversity (diatoms) and riparian carbon processing (cotton-strip assay), while minimizing losses in agricultural production potential. Regression models were developed to predict the target indicators for 489 segments of the Zwalm riparian corridor. For each riparian segment we evaluated alternative spatial allocations of different intensities of riparian reforestation to identify (1) trade-offs among the target indicators, (2) priority regions for reforestation, (3) required reforestation intensity. Riparian reforestation along the Zwalm could significantly improve biodiversity (up to +96% for EPT index) and ecosystem functioning indicators, but result in a strong trade-off with agricultural production. Our optimization analyses resulted in identification of the headwaters of the Zwalm as priority areas for reforestation actions, with tradeoffs further minimized through modest expansion of existing forest patches.