SETTING: Pulmonary tuberculosis (TB) patients reported in North Carolina (NC), USA, from 1996 to 2008 (inclusive). OBJECTIVE: To compare prevalence of cavitary TB as a surrogate marker for advanced disease in low-caseload counties with high-caseload counties. DESIGN: A multivariate log binomial regression model was used to estimate prevalence ratios (PRs) for cavitary TB. RESULTS: The proportion of TB cases in low-caseload counties vs. the total number of TB cases in NC over the study period increased from 10% in 1996 to 20% in 2008. After adjusting for human immunodeficiency virus (HIV) status, excess alcohol use and report year, patients in rural areas of low-caseload counties had greater prevalence (PR 1.40, 95%CI 1.19-1.64) of cavitary disease compared with patients from rural areas of high-caseload counties. The prevalence of cavitary TB did not differ between urban residents of high- or low-caseload counties (PR 1.00, 95%CI 0.86-1.16) after adjusting for HIV status, excess alcohol use and report year. DISCUSSION: TB patients in rural areas of low-caseload counties presented with more advanced TB disease compared with patients from urban and/or high-caseload counties. Barriers to timely recognition of TB in rural low-caseload settings must be considered in TB control programs.