Low copy number plasmids in bacteria require segregation for stable inheritance through celldivision. This is often achieved by a parABC locus, comprising an ATPase ParA, DNA binding protein ParB and a parC region, encoding ParB-binding sites. These minimal components space plasmids equally over the nucleoid, yet the underlying mechanism is not understood. Here we investigate a model where ParA-ATP can dynamically associate to the nucleoid and is hydrolyzed by plasmid-associated ParB, thereby creating nucleoid-bound, self-organizing ParA concentration gradients. We show mathematically that differences between competing ParA concentrations on either side of a plasmid can specify regular plasmid positioning. Such positioning can be achieved regardless of the exact mechanism of plasmid movement, including plasmid diffusion with ParA-mediated immobilization ordirected plasmid motion induced by ParB/parC-stimulated ParA structure disassembly. However, we find experimentally that parABC from Escherichia coli plasmid pB171increases plasmid mobility, inconsistent with diffusion/immobilization. Instead ourobservations favor directed plasmid motion. Our model predicts less oscillatory ParAdynamics than previously believed, a prediction we verify experimentally. We also show thatParA localization and plasmid positioning depend on the underlying nucleoid morphology, indicating that the chromosomal architecture constrains ParA structure formation. Ourdirected motion model unifies previously contradictory models for plasmid segregation and provides a robust mechanistic basis for self-organized plasmid spacing that may be widely applicable.