Group B streptococcus (GBS) remains the most common cause of early-onset sepsis in newborns. Laboratory gold-standard, broth culture methods are highly specific, but lack sensitivity. The aim of this study was to validate a nested-PCR and to determine whether residue volumes of urine samples obtained by non invasive, non sterile methods could be used to confirm neonatal GBS sepsis. The nested-PCR was performed with primers of the major GBS surface antigen. Unavailability of biological samples to perform life supporting exams, as well as others to elucidate the etiology of infections is a frequent problem concerning newborn patients. Nevertheless, we decided to include cases according to strict criteria: newborns had to present with signs and symptoms compatible with GBS infection; at least one of the following biological samples had to be sent for culture: blood, urine, or cerebrospinal fluid; availability of residue volumes of the samples sent for cultures, or of others collected on the day of hospitalization, prior to antibiotic therapy prescription, to be analyzed by PCR; favorable outcome after GBS empiric treatment. In only one newborn GBS infection was confirmed by cultures, while infection was only presumptive in the other three patients (they fulfilled inclusion criteria but were GBS-culture negative). From a total of 12 biological samples (5 blood, 3 CSF and 4 urine specimen), eight were tested by culture methods (2/8 were positive), and 8 were tested by PCR (7/8 were positive), and only 4 samples were simultaneously tested by both methods (1 positive by culture and 3 by PCR). In conclusion, although based on a restricted number of neonates and samples, our results suggest that the proposed nested-PCR might be used to diagnose GBS sepsis as it has successfully amplified the three types of biological samples analyzed (blood, urine and cerebrospinal fluid), and was more sensitive than culture methods as PCR in urine confirmed diagnosis in all four patients. Moreover, PCR has enabled us to use residue volumes of urine samples collected by non invasive, non sterile methods, what is technically adequate as GBS is not part of the normal urine flora, thus avoiding invasive procedures such as suprapubic bladder punction or transurethral catheterization. At the same time, the use of urine instead of blood samples could help preventing newborns blood spoliation.