Context .- All Food and Drug Administration-approved methods in the United States for human papillomavirus testing including the Hybrid Capture 2 human papillomavirus assay and the Roche cobas human papillomavirus test are approved for cytology specimens collected into ThinPrep media but not for specimens collected into SurePath solution. Objective .- To compare the performance of the Roche cobas and Hybrid Capture 2 tests for the detection of high-risk human papillomavirus using both ThinPrep and SurePath preparations as part of a validation study. Design .- One thousand three hundred seventy-one liquid-based cytology samples, including 1122 SurePath and 249 ThinPrep specimens, were tested for high-risk human papillomavirus DNA using the Roche cobas human papillomavirus test and the Hybrid Capture 2 human papillomavirus assay. For cases with discrepant results, confirmatory testing was performed using Linear Array human papillomavirus testing. Results .- One hundred and fifty-six (11.38%) and 184 (13.42%) of the 1371 specimens tested positive for high-risk human papillomavirus DNA using the Hybrid Capture 2 human papillomavirus assay and Roche cobas human papillomavirus assay, respectively. In addition, 1289 (94.0%) of 1371 specimens demonstrated concordant high-risk human papillomavirus results with a κ value of 0.72 (95% confidence interval, 065-0.78). There was no statistically significant difference in the percentage of positive high-risk human papillomavirus results between the 2 liquid-based preparations with either assay. Discordant results between the 2 assays were noted in 82 of 1371 cases (6%). Twenty-seven of 82 cases (32.9%) were Hybrid Capture 2 positive/Roche cobas negative and 55 of 82 cases (67.1%) were Roche cobas positive/Hybrid Capture 2 negative. Two of 20 Hybrid Capture 2-positive/Roche cobas-negative cases (10%) and 26 of 37 Roche cobas-positive/Hybrid Capture 2-negative cases (70%) tested positive for high-risk human papillomavirus by Linear Array. Conclusions .- Both assays showed good agreement and excellent specificity with either ThinPrep or SurePath preparations. The number of discordant results was relatively small. The performance of both assays was similar for ThinPrep specimens, but the Roche cobas test demonstrated higher sensitivity with SurePath specimens.