Human-associated microbial communities play important roles in health and disease. Antibiotic administration is arguably one of the most important modifiable determinants of the composition of the human microbiota. However, quantitatively modeling antibiotic use to account for its impact on microbial community dynamics presents a challenge. We used antibiotic therapy of chronic lung infection in persons with cystic fibrosis as a model system to assess the influence of key variables of therapy on measures of microbial community perturbation. We constructed multivariate linear mixed models with bacterial community diversity as the outcome measure and various scales of antibiotic weighting as predictors, while controlling for other variables. Antibiotic weighting consisted of three components: (i) dosing duration; (ii) timing of administration relative to sample collection; and (iii) antibiotic type and route of administration. Antibiotic weighting based on total dose and proximity to the time of sampling was most predictive of bacterial community change. Using this model to control for antibiotic use enabled the identification of other significant independent predictors of microbial community diversity such as dominant taxon, disease stage, and gender. Quantitative modeling of antibiotic use is critical in understanding the relationships between human microbiota and disease treatment and progression.