For a long time the candidate explanations were suction, friction, electrostatic charge and glue. Each fails a test. Geckos adhere in vacuum, which rules out suction. They adhere to molecularly smooth surfaces where friction is minimal. They leave no residue, so glue is out.
Measurements on single setae by Kellar Autumn and colleagues, published in 2000, showed forces consistent with van der Waals interactions: the setae split into spatula-shaped tips a couple of hundred nanometres wide, and the sheer number of contact points converts an individually negligible attraction into a substantial hold. The attachment is directional. Setae engage when dragged one way and release when peeled the other, which is why a gecko can run rather than being stuck fast, and why toes are hyperextended and peeled off from the tip.
Electrostatic contributions have been argued for in later work, so calling the mechanism exclusively van der Waals overstates the settled position. The house gecko is a good example of the trait in service of a niche: vertical walls near lights, where insects gather.