SummaryThis prospective study compared the diagnostic and predictive potential of sperm morphology assessments in a fertile vs. a subfertile population, evaluated in three different laboratories. The fertile population included 144 men who had recently fertilized their partners. As subfertile controls, 136 men with a history of subfertility for more than 12 months were used. All semen samples (280) were evaluated in three different centres in a blind fashion, without any patient information. The evaluation of sperm morphology was performed according to the criteria normally used in the different laboratories: WHO (1992) criteria for laboratory A, and Tygerberg strict criteria for laboratories B and C. Using ROC analysis, the predictive power of sperm morphology turned out to be different in the three laboratories (area under ROC curve: 69% for lab A, 72% for lab B and 78% for lab C). Using percentile 10 of the fertile population as the cut-off value for normality, we obtained the following results: 2, 1 and 5% for laboratories A, B and C, respectively. Using ROC analysis cut-off values with optimal speci®city and sensitivity were 6, 1 and 10%, respectively.Although our data highlight a reasonable predictive power of sperm morphology in centres using different or the same criteria, cut-off values for normality were different, even when the same criteria were applied. These results stress the importance of standardization in sperm morphology evaluation and the need for examining a reference population in estimating the real threshold value in different laboratories.