Our whole genome sequence (WGS) pipeline was assessed for accurate prediction of antimicrobial phenotypes. For 2316 invasive pneumococcal isolates recovered during 2015 we compared WGS pipeline data to broth dilution testing (
BDT) for 18 antimicrobials. For 11 antimicrobials categorical discrepancies were assigned when WGS-predicted MICs and
BDT MICs predicted different categorizations for susceptibility, intermediate resistance or resistance, ranging from 0.9% (
tetracycline) to 2.9% (
amoxicillin). For β-
lactam antibiotics, the occurrence of at least four-fold differences in MIC ranged from 0.2% (
meropenem) to 1.0% (
penicillin), although phenotypic retesting resolved 25%-78% of these discrepancies. Non-susceptibility to
penicillin, predicted by
penicillin-binding protein types, was 2.7% (non-
meningitis criteria) and 23.8% (
meningitis criteria). Other common resistance determinants included mef (475 isolates), ermB (191 isolates), ermB + mef (48 isolates), tetM (261 isolates) and cat (51 isolates). Additional accessory resistance genes (tetS, tet32, aphA-3, sat4) were rarely detected (one to three isolates). Rare core genome mutations conferring
erythromycin-resistance included a two-
codon rplD insertion (rplD69-KG-70) and the
23S rRNA A2061G substitution (six isolates). Intermediate
cotrimoxazole-resistance was associated with one or two
codon insertions within folP (238 isolates) or the folA I100L substitution (38 isolates), whereas full
cotrimoxazole-resistance was attributed to alterations in both genes (172 isolates). The two
levofloxacin-resistant isolates contained parC and/or gyrA mutations. Of 11 remaining isolates with moderately elevated MICs to both
ciprofloxacin and
levofloxacin, seven contained parC or gyrA mutations. The two
rifampin-resistant isolates contained rpoB mutations. WGS-based antimicrobial phenotype prediction was an informative alternative to
BDT for invasive pneumococci.