We evaluated the consistency of the A-DROP system for community-acquired pneumonia (CAP) and its outcomes, and developed a new severity classification of community-acquired pneumonia using nonlinear discriminant analysis. A total of 615 patients with CAP were enrolled between 2004 and 2009. A poor outcome was defined as patients requiring ventilation and/or death from CAP. We investigated the influence of prognostic factors on CAP severity and outcome using a logistic regression model to obtain the coefficient, and a contingency table. The optimal cutoff points for age and BUN were calculated from receiver-operating characteristic (ROC) curves. The influence of respiratory failure was approximately twice that of other prognostic factors. The optimal cutoff point for age was 83 years old, and that for BUN was 23mg/dl. We found inconsistencies in the equivalence of all prognostic factors and the addition-scoring method in predicting outcome. To ensure consistency between the A-DROP system and outcome, we believe that the weight of respiratory failure, threshold of classification, and cutoff points for age and BUN should be revised.