An improved NB/70K assay which can be used to reproducibly and accurately measure the NB/70K content of unextracted serum samples has been developed. Analysis of pretreatment serum samples has indicated that the NB/70K assay shows selectivity for ovarian cancer with respect to nonmalignant ovarian cysts and tumors as well as controls; if one considers 10 units of NB/70K as a significant level of antigen, then there are no false positives in the benign and control groups, while 10 of 21 in the ovarian cancer group have significant NB/70K levels. When NB/70K levels of ovarian cancer patients are compared to those of patients with non-ovarian malignant disease, only 4 of 23 patients with non-ovarian cancers had significantly elevated NB/70K levels. However, none of these 4 false positives had levels greater than 20 units.