BACKGROUND Human cytomegalovirus (HCMV) is the most common cause of viral intrauterine infection. Fetal damage is mostly linked to maternal primary infection. It is therefore important to differentiate primary from non-primary infection in pregnant females. IgM tests often used for this purpose are not reliable enough. OBJECTIVE To evaluate an HCMV-IgG urea-elution assay for its ability to distinguish primary from non-primary infection. In this assay, soaking the antigen-antibody complex with an urea containing solution frees antibodies with low avidity but has no influence on those with high avidity. An avidity index (AI) was calculated: AI = (OD with urea/OD without urea) x 100. METHODS HCMV-IgG avidity was measured on a single serum of 79 patients with past infection (pregnant women, graft recipients and blood donors) and of 63 patients (78 sera) with documented seroconversion (pregnant women and graft recipients). Sixty-one pregnant women positive or equivocal for HCMV-IgM but without a documented seroconversion were included in this study. RESULTS Most (72/79) of the patients with past infection had an AI > 65% and all but one had an AI > 50%. In pregnant women, in the case of a primary infection within the past 3 months, AI are usually (51/53) < 50% and never > 65%. Among the IgM positive pregnant women who lack a seroconversion history, 38 had AI > 65% suggestive of an infection that had occurred at least 3 months earlier, 11 had an AI in a grey area between 50 and 65% and 12 had an AI < 50%, suggestive of a recent primary infection. CONCLUSIONS In pregnant women, measurement of the IgG avidity may help to date a HCMV infection, an AI > 65% highly suggests a past infection while an AI < 50% corresponds to a recent primary infection.