The objective of this study was to assess the short-term test-retest variability of pressure-flow studies (PFS) in men with lower urinary tract symptoms (LUTS). By choosing a short interval between two consecutive PFS, but notably not performing two tests within a single session, both tests represent the same routine testing procedure. Eighty-nine patients with LUTS suggestive of bladder outlet obstruction or detrusor underactivity, who received PFS, were asked to undergo a second urodynamic evaluation within 4 weeks after the initial test. At both visits, specialized physicians performed the PFS. Obstruction was quantified using the Abrams-Griffiths number (AG number). Each patient was classified as obstructed, unobstructed, or equivocally obstructed according to the International Continence Society nomogram. As expected, no systematic difference was observed in AG number between the first and the second visit at the group level. There was, however, considerable variation at the individual level. The average within-patient standard deviation was 14 cm H(2)O. This finding shows that if a patient's AG number is 30 at the first visit, his true AG number can be any value between 30 +/- 1.96 * 14 = 3 to 57, owing to random variability alone. The average within-patient standard deviation did not differ significantly between sub-groups of obstruction and other variables such as patient age, symptom score, prostate volume, or residual volume. The variability appeared to increase slightly with an increased interval between visits. Of all patients, 39% changed at least one category of obstruction at the second visit and 3% changed from definitely obstructed to definitely unobstructed or vice versa. We conclude that PFS cannot stand the test of serving as a gold standard to identify bladder outlet obstruction in patients with LUTS.