Steady state human granulopoiesis was modeled by a convection-reaction differential equation of the Rubinow type for the bone marrow granulocyte precursors and an ordinary differential equation for the blood granulocytes. Measured values reported from several laboratories were used as sources for the model proliferation, maturation, and mobilization rates. Due to the large variability in the measured input data, four alternative models were constructed initially, each one with a specific combination of proliferation rate and maturation rate. They were all able to produce output values for the bone marrow neutrophil count and turnover rate close to accepted data, but neither of them could reproduce good values for the differential fractions of the neutrophil precursor stages. The model output was especially sensitive to changes in transit time in the mitotic relative to the postmitotic precursor compartments. When the net proliferation rate was modeled to optimize the bone marrow differential fractions according to published data, the total bone marrow neutrophil count would not fit with published data. However, a composite model optimizing differential fractions, bone marrow neutrophil count, and turnover rate yielded plausible output values and a reduced proliferation rate in the myelocyte stage. This result opens for a possibly substantial apoptosis rate at the myelocyte stage in accordance with results from earlier investigators. However, the result was based on a special choice of precursor transit times, taken from the literature. More precise data concerning granulocyte precursor cycle times, transit times, and differential fractions would radically improve the model's ability to clarify the role of apoptosis during granulocyte production and storage.