The in silico prediction of bacterial surface exposed proteins is of growing interest for the rational development of vaccines and in the study of bacteria-host relationships, whether pathogenic or host beneficial. This interest is driven by the increase in the use of DNA sequencing as a major tool in the early characterization of pathogenic bacteria and, more recently, even of complex ecosystems at the host-environment interface in metagenomics approaches. Current protein localization protocols are not suited to this prediction task as they ignore the potential surface exposition of many membrane-associated proteins. Therefore, we developed a new flow scheme, SurfG+, for the processing of protein sequence data with the particular aim of identification of potentially surface exposed (PSE) proteins from Gram-positive bacteria, which was validated for Streptococcus pyogenes. The results of an exploratory case study on closely related lactobacilli of the acidophilus group suggest that the yogurt bacterium Lactobacillus delbrueckii ssp. bulgaricus (L. bulgaricus) dedicates a relatively important fraction of its coding capacity to secreted proteins, while the probiotic gastrointestinal (GI) tract bacteria L. johnsonii and L. gasseri appear to encode a larger variety of PSE proteins, that may play a role in the interaction with the host.