Immunological correlates of protection-values of immunological assays associated with protection from disease-play a central role in vaccines research.Although in reality the relationship is likely continuous,increasing assay values corresponding to increased protection,in practice the use of threshold protective levels is ubiquitous.The situation may be likened to general clinical medicine,where threshold reference levels have been established for many common laboratory tests,such as in adults ＜200 mg/dL for cholesterol,0.7-1.5 mg/dL for serum creatinine,and 4-10.8×103/μL for white blood cell count.Yet limited quantitative methods exist for inferring protective thresholds from data on immunological assay values and subsequent disease occurrence.Standard methods are often unsuitable,since in most study designs a large number of unprotected subjects do not develop disease,in addition to those who are protected.This research examines eight threshold models and methods in terms of their suitability for estimating threshold immumological correlates of protection,and develops tests and measures of discrimination and goodness-of-fit for evaluating thresholds so found.Results are illustrated using fifteen datasets.