We report the results of the first three trials of an external quality control (EQC) programme performed in 71 laboratories executing semen analysis in Tuscany Region (Italy). At the end of the second trial, participants were invited to attend a teaching course illustrating and inviting to adhere to procedures recommended by WHO (V edition). Results of the first three trials of the EQC documented a huge variability in the procedures and the results. The highest variability was found for morphology (CV above 80% for all the trials), followed by count (CV of about 60% for all the trials) and motility (CV below 30% for all the trials). When results of sperm count and morphology were divided according to the used method, mean CV values did not show significant differences. CV for morphology dropped significantly at the third trial for most methods, indicating the usefulness of the teaching course for morphology assessment. Conversely, no differences were observed after the course for motility and for most methods to evaluate count, although CV values were lower at the second and third trial for the laboratories using the Burker cytometer. When results were divided according to tertiles of activity, the lowest mean bias values (difference between each laboratory result and the median value of the results) for count and morphology were observed for laboratories in the third tertile (performing over 200 semen analysis/year). Of interest, mean bias values for concentration dropped significantly at the third trial for low activity laboratories. In conclusion, lack of agreement of results of semen analysis in Tuscany is mainly because of the activity and the experience of the laboratory. Our study points out the importance of participating in EQC programmes and periodical teaching courses as well as the use of WHO recommended standardized procedures to increase precision and to allow the use of WHO reference values.