We evaluated the ability of repeated measurements of circulating immune complexes (CIC) to predict for tumor recurrence in 130 patients with malignant melanoma. Twenty-two patients had level 2, 45 had level 3, and 51 had level 4/5, stage I disease in remission at the start of monitoring, while 12 had stage II disease. The polyethylene glycol precipitation assay was used for serial studies, based on an initial comparative evaluation with the Clq-binding and Raji assays. The study averaged 22 ± 11 months (6–43 months) and an average of 22 ± 5.3 assays were performed per patient (range 3–36), with a follow-up of 4 years. CIC were present in sera in recurrent, irregular ‘bursts’ of activity. Serial measurements doubled the incidence of CIC compared to single determinations. Only 23% of these bursts of activity were clearly related temporarily to documented recurrences, while 34% occurred with treatment events such as surgery or immunotherapy, and 42% occurred without correlation to either recurrence or treatment. CIC activity was greater and more closely related to recurrence in high-risk stage I (level 4,5) and stage II patients. Whether analyzed as positive sera or as bursts of elevated CIC activity, CIC assays predicted for recurrence at the 5% significance level. The assay was highly sensitive (97%), but with poor specificity (21%) with many false positives (79%). The assay was helpful at ruling out recurrences (95%), but poor at ruling them in (29%). The advantage was seen only in high-risk stage I and II patients, and there was no advantage to serial assays over random single determinations. Although generally, CIC in the sera of melanoma patients were found to predict for recurrence, the use of serial CIC measures monitoring of individual patients cannot be recommended.