The pharmacodynamic potency of a therapeutic cytokine interacting with a cell-surface receptor can be attributed primarily to three central properties: [1] cytokine/receptor binding affinity, [2] cytokine/receptor endocytic trafficking dynamics, and [3] cytokine/ receptor signaling. Thus, engineering novel or second-generation cytokines requires an understanding of the contribution of each of these to the overall cell response. We describe here an efficient method toward this goal in demonstrated application to the clinically important cytokine granulocyte colony-stimulating factor (GCSF) with a chemical analogue and a number of genetic mutants. Using a combination of simple receptor-binding and dose-response proliferation assays we construct an appropriately scaled plot of relative mitogenic potency versus ligand concentration normalized by binding affinity. Analysis of binding and proliferation data in this manner conveniently indicates which of the cytokine properties-binding, trafficking, and/or signaling-are contributing substantially to altered potency effects. For the GCSF analogues studied here, two point mutations as well as a poly(ethylene glycol) chemical conjugate were found to have increased potencies despite comparable or slightly lower affinities, and trafficking was predicted to be the responsible mechanism. A third point mutant exhibiting comparable binding affinity but reduced potency was predicted to have largely unchanged trafficking properties. Surprisingly, another mutant possessing an order-of-magnitude weaker binding affinity displayed enhanced potency, and increased ligand half-life was predicted to be responsible for this net beneficial effect. Each of these predictions was successfully demonstrated by subsequent measurements of depletion of these five analogues from cell culture medium. Thus, for the GCSF system we find that ligand trafficking dynamics can play a major role in regulating mitogenic potency. Our results demonstrate that cytokine analogues can exhibit pharmacodynamic behaviors across a diverse spectrum of "binding-potency space" and that our analysis through normalization can efficiently elucidate hypotheses for the underlying mechanisms for further dedicated testing. We have also extended the Black-Leff model of pharmacological agonism to include trafficking effects along with binding and signaling, and this model provides a framework for parsing the effects of these factors on pharmacodynamic potency.