Penile pharmacotesting (PPT) with alprostadil (PGE1) represents the most common diagnostic approach to male erectile dysfunction (ED). A positive response - i.e. normal erectile rigidity of sustained duration - is presumed to exclude venous or arterial pathology with enough accuracy. To test this assumption we compared PPT vs. flowmetric results obtained by colour-duplex Doppler ultrasound (CDDU) in patients (pts) undergoing diagnostic evaluation for ED under conditions of maximal cavernous relaxation. A total of 195 non-consecutive impotent pts were diagnosed after dynamic CDDU as non-vasculogenic (NOR), or having arteriogenic (AR), veno-occlusive (VO) or mixed (MX) ED. Maximal erection obtained after PPT was scored as: type-1 (full tumescence - no sustained rigidity, angle on the abdominal plane >90 degrees), type-2 (sustained partial erection, valid for intromission, angle=90 degrees) and type-3 (sustained full erection, angle <90 degrees). Comparing PPT with flowmetric results, we found that a type-3 response had 20% false negative diagnosis of NOR (17% of AR- and 3% of VO- and MX-ED, respectively), while a type-2 response had 63% false negative diagnosis (20% of AR, 37% of VO- and 6% MX-ED, respectively). Type-1 response was associated with the presence of VO dysfunction in 99% of cases. These data suggest that a positive response to PPT (type-2 and type-3) assessed by the visual rating of erection is associated with both arterial (up to 20%) and/or VO (up to 43%) ED, as detected by CDDU. We conclude that PPT alone is a misleading diagnostic test to exclude vascular ED and that dynamic CDDU should be offered to pts investigated for male ED.