Abstract Study question Guidelines suggest that one semen analysis is sufficient during the diagnostic work-up of an infertile man in the case of normality as for WHO criteria. Summary answer We investigated the rate and the clinical features of men with abnormal sperm parameters at a second test, after a normal first semen analysis. What is known already A second test is recommended when the first semen analysis depicted abnormal sperm parameters. Study design, size, duration Complete demographic, clinical and laboratory data from 1358 consecutive primary infertile men (infertility as for WHO definition) were analysed. Health-significant comorbidities were scored with the Charlson Comorbidity Index (CCI). Serum hormones were measured in every patient. Patients underwent two consecutive semen analyses at the same laboratory, which followed 2010 WHO reference criteria. Participants/materials, setting, methods Descriptive statistics and logistic regression models tested the association between clinical variables and semen parameters. Receiver operative characteristic (ROC) curves were used to assess the relationship between clinical variables and to create a composite risk score for pathological sperm parameters at a second test. Main results and the role of chance At first analysis, 212 (15.6%) infertile men had normal semen parameters. Of 212, 87 (41.0%) had a second normal semen analysis, while 80 (37.7%), 35 (16.5%) and 10 (4.7%) men showed 1, 2 and 3 pathological sperm parameters at second test. Men with a pathological second semen analysis had higher CCI scores (p < 0.001), smaller testicular volume (p < 0.001) and higher FSH values (p < 0.01) than those with normal second samples. Overall, despite being within normal ranges, sperm concentration was lower [34 (23–57) vs. 62 (35–94); p < 0.01] in men with an abnormal second sample compared to those with confirmed normality. At multivariable logistic regression analysis, smaller testicular volume (OR 0.9, p = 0.03), FSH (OR 1.2, p < 0.01), and lower sperm concentration (OR 0.9, p < 0.01) were associated with pathological second semen analyses, after accounting for age and CCI. ROC curves showed that testicular volume <15 ml, FSH values >6 mUI/ml and sperm concentration <40 mil/ml had good predictive ability for pathologic second sperm parameters (all AUC >0.8). Considering 1-point for each of the previous variables, the chances of a pathological second analysis increased from 38.8% to 74.6%, 77.3% and 100% among patients with risk scores of 0, 1, 2 and 3, respectively (p < 0.001). Limitations, reasons for caution It is a retrospective analysis at a single, tertiary-referral academic centre, thus raising the possibility of selection biases. In spite of this, all patients have been consistently analysed over time with a rigorous follow-up, thus limiting potential heterogeneity in terms of data reporting. Wider implications of the findings: Approximately 60% of infertile men with a normal semen analysis depicted a pathological second test. Smaller testicles, higher FSH, lower sperm concentrations were independently associated with a pathologic second test. These features could be useful to identify those infertile men with a normal semen analysis who deserve a second test. Trial registration number Not applicable