The paper presents a method called LipoP for predicting lipoprotein signal peptides in Gram-negative bacteria. The method uses a hidden Markov model (HMM) to distinguish between lipoproteins, SPaseI-cleaved proteins, cytoplasmic proteins, and transmembrane proteins. The HMM achieved a 96.8% accuracy rate in predicting lipoproteins with only 0.3% false positives in a set of SPaseI-cleaved, cytoplasmic, and transmembrane proteins. The method was also tested on a Gram-positive test set, achieving 92.9% accuracy. The HMM was further evaluated on genome searches for 12 Gram-negative and one Gram-positive genome, with results compared to experimental data. A neural network-based predictor was developed for comparison and yielded similar results. The LipoP server is available at www.cbs.dtu.dk/services/LipoP/. The paper discusses the characteristics of lipoprotein signal peptides and the differences between Gram-negative and Gram-positive lipoproteins, providing a detailed analysis of the methods used and their performance.The paper presents a method called LipoP for predicting lipoprotein signal peptides in Gram-negative bacteria. The method uses a hidden Markov model (HMM) to distinguish between lipoproteins, SPaseI-cleaved proteins, cytoplasmic proteins, and transmembrane proteins. The HMM achieved a 96.8% accuracy rate in predicting lipoproteins with only 0.3% false positives in a set of SPaseI-cleaved, cytoplasmic, and transmembrane proteins. The method was also tested on a Gram-positive test set, achieving 92.9% accuracy. The HMM was further evaluated on genome searches for 12 Gram-negative and one Gram-positive genome, with results compared to experimental data. A neural network-based predictor was developed for comparison and yielded similar results. The LipoP server is available at www.cbs.dtu.dk/services/LipoP/. The paper discusses the characteristics of lipoprotein signal peptides and the differences between Gram-negative and Gram-positive lipoproteins, providing a detailed analysis of the methods used and their performance.