Research questionDoes the choice of sperm-counting chamber affect the proportion of samples generating results with an erroneous interpretation? DesignLaboratories in an external quality assurance programme were sent 141 semen samples over a 12-year period and asked to return the sperm concentration and whether or not the result was abnormal. Only those using 5th edition of the World Health Organization manual (WHO5) interpretation criteria were included. Submissions from specialist fertility laboratories were used to calculate assigned values for each sample. Laboratory50 values determined the sperm concentration at which the laboratories reported a majority transition from abnormal to normal interpretations, i.e. the tipping point, which should coincide with the lower reference limit. ResultsThe median and range of bias from the assigned values of each sample were determined for the Makler (–3.3%; –88.6% to +332.8%), haemocytometer (10.6%; –93.3% to +645.5%), Kova (+65.3%; –71.7% to +581.8%) and Vetriplast (+72.4%; –100.0% to +709.1) chambers. Laboratory50 values for the Makler (17.3 × 106/ml), haemocytometer (13.6 × 106/ml), Kova (10.0 × 106/ml) and Vetriplast chambers (8.8 × 106/ml) reflected the under- and overestimation of the chambers and confirmed a shift in the adjusted lower reference limit then used. The proportion of laboratories reporting erroneous interpretations of the four chambers for oligozoospermic samples were 10.9%, 15.1.%, 40.1% and 44.0%, respectively, and rose as the adjusted lower reference limit decreased. ConclusionsThe between-laboratory and within-sample variation for all the chambers was high and remains a concern. The main impact of an increasing bias of the chambers was a lowering of the laboratory50 tipping point, resulting in an under-reporting of abnormal semen samples.