Fine-structure-specific predissociation Linewidths have been calculated for the v = 0-18, N = 0-40 levels of the B (3) Sigma(u)(-) state of O-16(2) using a predissociation model which takes into account spin-orbit interactions between the B (3) Sigma(u)(-) state and the 1 (1) Pi(u), 1 (3) Pi(u), 1 (5) Pi(u), and 2 (3) Sigma(u)(+) states, and spin-electronic and L-uncoupling interactions between the B (3) Sigma(u)(-) and 1 (3) Pi(u) states. The model parameters were optimized by comparing the calculated widths with existing measurements of fine-structure Linewidth for v = 0, 14-18, and new determinations for v = 1, 2, 5, 7, 9-13 which have been obtained from existing cross sections for the (v,0) and (v,1) Schumann-Runge bands using a fitting procedure which assumes unequal Linewidths for the triplet fine-structure components. The L-uncoupling interaction is found to be important for all vibrational levels in determining the width ratios for the fine-structure components, even at quite low rotational levels. The calculated linewidths vary from 0.03 (for v = 16, N = 30, F-1) to 4.8 cm(-1) (for v = 4, N = 40, F-3), and good agreement is found between the measurements and the model calculations for most vibrational levels. The calculations are expected to be useful in the construction of state-of-the-art models describing the photochemistry of the middle atmosphere. Greatly improved predissociation model parameters, in good agreement with ab initio calculations, have been obtained for the (3) Pi(u) and (1) Pi(u) states, while the (3) Sigma(u)(+) parameters remain the most uncertain. The model parameters resulting in the best fit to the widths also give good agreement with observed perturbations in the Schumann-Runge band origins. It is found that the actual separations between triplet fine-structure levels consistently exceed values predicted from grenerally accepted spectroscopic constants for the B (3) Sigma(u)(-) state, suggesting that a reevaluation of those constants may be in order.