Polymicrobial bronchopulmonary infections in cystic fibrosis (CF) cause progressive lung damage and death. Although the arrival of Pseudomonas aeruginosa often heralds a more rapid rate of pulmonary decline, there is significant inter-individual variation in the rate of decline, the causes of which remain poorly understood. By coupling culture-independent methods with ecological analyses, we discovered correlations between bacterial community profiles and clinical disease markers in respiratory tracts of 45 children with CF. Bacterial community complexity was inversely correlated with patient age, presence of P. aeruginosa and antibiotic exposure, and was related to CF genotype. Strikingly, bacterial communities lacking P. aeruginosa were much more similar to each other than were those containing P. aeruginosa, regardless of antibiotic exposure. This suggests that community composition might be a better predictor of disease progression than the presence of P. aeruginosa alone and deserves further study.