A systematic review was conducted of published studies, which related post-coital test (PCT) results to pregnancy, to evaluate the predictive power of the test and determine a definition of abnormality that yielded the best test properties. Well defined test results and valid outcome data for women attending fertility clinics were present in 11 out of 53 published reports relating to a total of 3093 women. The predictive values of normal and abnormal PCT were 0.37-0.92 and 0.58-0.85 respectively. Sensitivity was 0.10-0.90 and specificity 0.30-0.97. Likelihood ratios for normal and abnormal PCT were 0.77 and 1.85 respectively. When criteria for normality were set at a lower number of motile spermatozoa per field, sensitivity decreased with an increase in specificity and likelihood ratio for an abnormal test result. The discriminating ability of the PCT is poor, and altering definitions of normality hardly enhances its predictive power. As long as the value of the PCT for the assessment and treatment of so-called 'cervical factor infertility' remains unclear, a cut-off point with high specificity and a high likelihood ratio for an abnormal test result is recommended. Hence, an abnormal PCT is best defined as less than one motile spermatozoon per high power field.