During the epizootic of highly pathogenic avian influenza A(H7N7) in 2003 in The Netherlands, RT-PCR and culture confirmed infection was detected in 89 persons who were ill. A modified hemagglutination inhibition (HI) test using horse erythrocytes and 2 hemagglutinating units of virus was applied to assess retrospectively the extent of human (subclinical) infection. Validation of the HI-test with sera from 34 RT-PCR and culture confirmed A(H7) infected persons and sera from 100 persons from a human influenza vaccine trial in autumn 2002 showed that this HI-test had a sensitivity of 85% and a specificity of 100% when using a cut-off titer of > or =10. Using this cut-off value, A(H7) specific antibodies were detected in 49% of 508 persons exposed to poultry and in 64% of 63 persons exposed to A(H7) infected persons. Correlation of seropositivity with the occurrence of eye symptoms in exposed persons who had not received antiviral prophylaxis and of reduced seropositivity with taking antiviral prophylaxis provided further evidence that the A(H7) HI antibody titers were real. In conclusion, by applying an HI-test using horse erythrocytes human antibodies against the avian A(H7N7) virus were detected with high sensitivity and specificity in an unexpectedly high proportion of exposed persons.