peer reviewed ; Background: : GBS cause severe infections in neonates, pregnant women and other adults. Empiric therapy is usually started before susceptibility results are available. Early neonatal diseases can be prevented with intrapartum antibiotic prophylaxis based on accurate susceptibility surveillance data. A previous Belgian study showed an increase of 3 to 10 % R to erythromycin (EM) through the 1990s. Methods: 187 GBS isolates consecutively received at the reference laboratory between 2001 to March 2003 were from 73 neonates (52 early-onset and 21 late-onset diseases), 52 adults and 62 from pregnant women’s vagina. MICs of penicillin (PG), EM, clindamycin (CM) and gentamicin (GM) were determined with Etest. PG MBCs were also determined by inactivating the drug in MIC plates using betalactamase. EM resistant (R) isolates were tested by the CM + EM double disk to determine macrolide R phenotypes. Results: All strains were susceptible (S) to PG and no tolerance was observed with MBCs falling within 2 dilutions of MICs. 19.2% of isolates were R to EM, with significantly more R isolates from adults (30.8%; p <0.01) and serotype V (46.8%; p <0,001). 80% had the MLSB phenotype (R to EM and CM), 16 were constitutive and 12 inducible. The M phenotype (R to EM and S to CM) was seen in 7 (20%) of isolates. Less than 10% of isolates were inhibited by GM MIC of <=64 mg/L, 83.6% by 128-256 mg/L and 2.9% by >/=512 mg/L. Non typable strains were more R to GM (p <0.01). Conclusions: 1) PG remained active against all isolates and no tolerance was seen. 2) Prevalence of R to macrolides had increased since 1999, particularly in adult isolates and serotype V. 3) Intermediate to high level R to GM was seen and potential synergy of PG + GM should be investigated. 4) R surveillance is mandatory to guide prophylaxis and treatment of serious GBS infections.