Objectives Our objective was to evaluate and exploit a whole genome sequence (WGS) bioinformatics pipeline for predicting antimicrobial resistance and capsular serotypes from invasive group B streptococci (iGBS). Methods For 1975 iGBS recovered during 2015 from CDC's Active Bacterial Core surveillance, we compared pipeline predictions with broth dilution testing. Fifty-six isolates from earlier surveillance were included for testing β-lactams. Conventional serotyping was compared to WGS-based assignments for 302 isolates. Results All 28 isolates with reduced susceptibility to β-lactam antibiotics harboured one of 19 rare PBP2x types. Resistances to erythromycin/clindamycin (808/1975 isolates, 41.0%), erythromycin (235/1975, 11.9%) and lincosamide/streptogramin A/pleuromutilins (56/1975, 2.8%) were predicted by the presence of erm-methylase, mef and lsa determinants, respectively (41 of 56 lsa gene-positive isolates also contained lnu, erm and/or mef genes). Presence of both erm and lsa determinants (25 isolates) predicted non-susceptibility to quinupristin/dalfopristin. Most isolates (1680/1975, 85.1%) were tet gene-positive, although 41/1565 (2.6%) tetM-positive isolates were tetracycline-susceptible. All 53 fluoroquinolone-resistant isolates contained ParC and/or GyrA substitutions. Resistances to rifampin (eight isolates), trimethoprim, chloramphenicol and vancomycin (two isolates each) were predicted by the pipeline. Resistance to macrolides/lincosamides without pipeline prediction was rare and correlated to divergent resistance genes or rRNA A2062G substitution. A selection of 267 isolates assigned WGS-based serotypes were also conventionally serotyped. Of these, 246 (92.1%) were in agreement, with the remaining 21 (7.8%) conventionally non-serotypeable. For 32 of 1975 isolates (1.6%), WGS-based serotypes could not be assigned. Conclusion The WGS-based assignment of iGBS resistance features and serotypes is an accurate substitute for phenotypic testing.