The relationships between processing conditions and ultimate membrane properties that exist in the spinning of hollow fiber membranes are not fully understood. To address the need for a widely applicable fiber spinning model, the present work builds on the thin filament analysis (TFA) model of the fiber spinning process and predicts the variation of fiber size, velocity, and temperature in the draw zone, including the effects of surface tension. By modeling concentration gradients in the draw zone, this model also predicts both the extent of anisotropy (portion of membrane thickness containing a pore size gradient) and degree of anisotropy (indicative of the pore size range across the membrane thickness). For typical hollow fiber membrane spinning, predicted membrane extent of anisotropy is most sensitive to the diffusion coefficient. The predicted degree of anisotropy is most sensitive to the viscosity activation energy, affected significantly by the diffusion coefficient and spinning temperature, and affected slightly by the air gap length. Trends in the model-predicted and experimental measurements of final fiber outer diameter, ratio of outer to inner diameter (OD/ID), and extent of anisotropy agree. While model sensitivity studies show that surface tension can reduce the rate of fiber elongation, it has little effect on the predicted degree and extent of anisotropy. Furthermore, this analysis allows prediction of counter-intuitive concentration changes associated with temperature dependence of the diluent vapor pressure: for some cases, the degree of anisotropy can increase and then decrease along the draw zone during spinning. Reducing the dependence of vapor pressure on temperature eliminates this phenomenon. (C) 2009 Elsevier B.V. All rights reserved.