This paper presents an improved method for predicting linear B-cell epitopes, called BepiPred. The method combines a hidden Markov model (HMM) with a propensity scale method. The study used three data sets of proteins with linear B-cell epitope annotations: the Pellequer data set, the AntiJen data set, and the HIV data set. The methods were validated on an independent data set not used for training or optimization. The performance of the methods was measured using ROC-curves. The best single method for predicting linear B-cell epitopes is the HMM. Combining the HMM with the best propensity scale methods (Parker and Levitt) resulted in the BepiPred method, which performed significantly better than other methods on the validation data set. The BepiPred method is publicly available at http://www.cbs.dtu.dk/services/BepiPred. The study found that the best propensity scale method for predicting linear B-cell epitopes is the one by Levitt, followed by the Parker method. The HMM method performed better than the best propensity scale methods on the Pellequer data set. When combined with the Parker method, the HMM method produced the BepiPred method, which performed best on the validation data set. The study also found that the best methods for predicting linear B-cell epitopes have a relatively low sensitivity. To improve sensitivity, the threshold can be lowered, but this would also reduce specificity. The study suggests that combining prediction curves could help reduce over-predictions and improve B-cell epitope prediction methods. The study concludes that BepiPred is a novel method for predicting linear B-cell epitopes, combining the predictions of an HMM and a propensity scale method. The method was tested using non-parametric ROC-curves and validated on an independent data set. BepiPred showed the highest prediction accuracy on the test data set and performed significantly better than other methods on the validation data set. The method is publicly available and can be used for predicting linear B-cell epitopes in pathogenic organisms.This paper presents an improved method for predicting linear B-cell epitopes, called BepiPred. The method combines a hidden Markov model (HMM) with a propensity scale method. The study used three data sets of proteins with linear B-cell epitope annotations: the Pellequer data set, the AntiJen data set, and the HIV data set. The methods were validated on an independent data set not used for training or optimization. The performance of the methods was measured using ROC-curves. The best single method for predicting linear B-cell epitopes is the HMM. Combining the HMM with the best propensity scale methods (Parker and Levitt) resulted in the BepiPred method, which performed significantly better than other methods on the validation data set. The BepiPred method is publicly available at http://www.cbs.dtu.dk/services/BepiPred. The study found that the best propensity scale method for predicting linear B-cell epitopes is the one by Levitt, followed by the Parker method. The HMM method performed better than the best propensity scale methods on the Pellequer data set. When combined with the Parker method, the HMM method produced the BepiPred method, which performed best on the validation data set. The study also found that the best methods for predicting linear B-cell epitopes have a relatively low sensitivity. To improve sensitivity, the threshold can be lowered, but this would also reduce specificity. The study suggests that combining prediction curves could help reduce over-predictions and improve B-cell epitope prediction methods. The study concludes that BepiPred is a novel method for predicting linear B-cell epitopes, combining the predictions of an HMM and a propensity scale method. The method was tested using non-parametric ROC-curves and validated on an independent data set. BepiPred showed the highest prediction accuracy on the test data set and performed significantly better than other methods on the validation data set. The method is publicly available and can be used for predicting linear B-cell epitopes in pathogenic organisms.