The time course of recovery in the Guillain-Barré syndrome is known to vary widely, but factors associated with differences have not been previously defined. In this study we used multivariate analysis to identify such factors and to determine whether the presence or absence of specific factors would influence treatment decisions, particularly the use of plasmapheresis. Data from 245 patients randomized into conventional and plasmapheresis arms were used to assess the time to walk independently (Grade 2), the time to improve one grade, and the percentage improved at 4 weeks. Individually, many factors were associated with outcome. In the multivariate analysis, four factors correlate with poorer outcomes: mean amplitude of compound muscle action potential on stimulating distally of 20% of normal or less, older age, time from onset of disease of 7 days or less, and need for ventilatory support. The most powerful predictor of outcome was the abnormal mean amplitude of compound muscle action potential on stimulating distally. Plasmapheresis, the only variable the physician can influence, has a beneficial effect over and above any or all of these factors. The plasmapheresis patients on continuous flow machines had better outcomes than those on intermittent flow machines. From these data, tables of expected outcome probabilities have been developed.