In an ongoing effort to design an efficacious, cost-effective ovarian cancer screening method, the existing tests, CA 125 and transvaginal sonography, are being optimized and combined in a multimodal strategy, and new promising serum markers, such as mesothelin and HE4, are being developed and evaluated. Detection has been found to improve when multiple serum markers are used in a longitudinal logarithm. The parametric empirical Bayes approach improves screening algorithms by capturing the stability of markers over time in a heterogeneous population. It also has relatively simple extensions to multiple markers. The evaluation of markers increasingly accounts for characteristics of a woman that may affect her marker levels and accounts for the cancer's characteristics, histology, and grade. Receiver operating characteristic curves are helpful for evaluation because they relate a marker's sensitivity to the specificity at which it operates. Large, well-designed randomized controlled trials are under way to gauge the performance of existing screening methods.