OBJECTIVE Alpha-fetoprotein (AFP) is an important tumor marker for non-seminomatous germ cell tumors (NSGCTs) that greatly affects diagnosis and the evaluations of therapy and therapeutic policy. However, it is sometimes very difficult to make the distinction between tumors and falsely elevated AFP levels due to benign liver disease. We assessed the usefulness of lectin-reactive alpha-fetoprotein (AFP-L3%), which has been reported to be superior to total AFP in both sensitivity and specificity in hepatocellular carcinomas, for the evaluation of predictions of clinical courses after the treatment of NSGCTs. METHODS Frozen sera of 25 tumor-bearing patients with testicular cancers whose AFP levels were 5.0 ng/ml or higher were used. The total AFP levels and the ratio of L3 fraction to total AFP (AFP-L3%) were measured by liquid-phase binding assay (LBA). RESULTS The total AFP levels were 6.3-14 907 ng/ml (median: 105.9 ng/ml). The median AFP-L3% was 69.9% (range;: 1.1-88.1%). Except for one patient, 24 patients (96.0%) with evident disease showed high levels of AFP-L3% of >50%, regardless of their total AFP levels. In nine patients whose sera were sequentially measured, AFP-L3% was considered highly useful for the detection of residual tumors (n = 2) and recurrence (n = 1) and for the exclusion of false-positive cases (n = 1). CONCLUSIONS When the total AFP level increases slightly (e.g. to 5-20 ng/ml), a measurement of AFP-L3% may provide additional useful information for monitoring NSGCT patients and in distinguishing falsely elevated AFP.