Many theorists think that implicatures are not only cancellable, but eo ipso deniable: the speaker can deny having meant whatever she did not explicitly state. The advantage of implicit communication is thought to be avoiding commitment and shifting responsibility onto the hearer. Recently some theorists have pointed out that actually many implicatures can hardly be denied. Still the conviction persists that it is somehow in the nature of implicatures to be deniable. I will argue that though many implicatures certainly are deniable, it is not in their nature to be so.