To make feasible future clinical trials with new-generation human papillomavirus (HPV) vaccines, novel virological surrogate endpoints of progressive disease have been proposed, including high-risk HPV (HR-HPV) persistence for six months (6M+) or 12 months (12M+). The risk estimates (relative risks [RRs]) of these 'virological endpoints' are influenced by several variables, not yet validated adequately. We compared the impact of three referent groups: (i) HPV-negative, (ii) HPV-transient, (iii) HPV-mixed outcome on the risk estimates for 6M+ or 12M+ HR-HPV persistence as predictors of progressive disease. Generalized estimating equation models were used to estimate the strength of 6M+ and 12M+ HR-HPV persistence with disease progression to squamous intraepithelial lesions (SILs), cervical intraepithelial neoplasia (CIN) grade 1+, CIN2+, CIN/SIL endpoints, comparing three optional reference categories (i)-(iii) in a prospective sub-cohort of 1865 women from the combined New Independent States of the Former Soviet Union (NIS) and Latin American Screening (LAMS) studies cohort (n = 15,301). The RRs of these viral endpoints as predictors of progressive disease are affected by the length of viral persistence (6M+ or 12M+) and the surrogate endpoint (SIL, CIN1, CIN2, CIN/SIL). Most dramatic is the effect of the referent group used in risk estimates, with the HPV-negative referent group giving the highest and most consistent RRs for both 6M+ and 12M+ viral persistence, irrespective of which surrogate is used. In addition to deciding on whether to use 6M+ or 12M+ persistence criteria, and cytological, histological or combined surrogate endpoints, one should adopt the HPV-negative referent group as the gold standard in all future studies using viral persistence as the surrogate endpoint of progressive disease.