36Resistance to amoxicillin-clavulanate, a widely used beta-lactam/beta-lactamase inhibitor 37 combination antibiotic, is rising globally, yet susceptibility testing remains challenging. To test 38 whether whole-genome sequencing (WGS) could provide a more reliable assessment of 39 susceptibility than traditional methods, we predicted resistance from WGS for 976 E. coli 40 bloodstream infection isolates from Oxfordshire, UK, comparing against phenotypes from the 41 BD Phoenix (calibrated against EUCAST guidelines). 339/976 (35%) isolates were amoxicillin-42 clavulanate resistant. Predictions based solely on beta-lactamase presence/absence performed 43 poorly (sensitivity 23% (78/339)) but improved when genetic features associated with 44 penicillinase hyper-production (e.g. promoter mutations, copy number estimates) were 45 considered (sensitivity 82% (277/339); p<0.0001). Most discrepancies occurred in isolates with 46 peri-breakpoint MICs. We investigated two potential causes; the phenotypic reference and the 47 binary resistant/susceptible classification. We performed reference standard, replicated 48 phenotyping in a random stratified subsample of 261/976 (27%) isolates using agar dilution, 49 following both EUCAST and CLSI guidelines, which use different clavulanate concentrations. 50 As well as disagreeing with each other, neither agar dilution phenotype aligned perfectly with 51 genetic features. A random-effects model investigating associations between genetic features and 52 MICs showed that some genetic features had small, variable and additive effects, resulting in 53 variable resistance classification. Using model fixed-effects to predict MICs for the non-agar 54 dilution isolates, predicted MICs were in essential agreement (±1 doubling dilution) with 55 observed (BD Phoenix) MICs for 691/715 (97%) isolates. This suggests amoxicillin-clavulanate 56 resistance in E. coli is quantitative, rather than qualitative, explaining the poorly reproducible 57 4 binary (resistant/susceptible) phenotypes and suboptimal concordance between different 58 phenotypic methods and with WGS-based predictions. 59 60 61Rising amoxicillin-clavulanate resistance in E. coli is a major healthcare challenge, with 62 increasing incidence of resistant bloodstream infections (BSI)(1) threatening its utility as the 63 most commonly used antibiotic in Europe.(2) Consequently, many hospitals are considering 64 broadening their first-line empiric antibiotics for common infections. However, significant 65 uncertainty is created by observed differences between the two main assays for amoxicillin-66 clavulanate susceptibility in the classification of clinical samples.(3) These differences are so 67 large that increasing amoxicillin-clavulanate resistance was suggested to be primarily due to 68 laboratories switching from US Clinical Laboratory Standards Institute (CLSI) to European 69 Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines.(4) Recent work,(5) 70 however, suggests that changes in laboratory protocols ar...