For the generalist spider Cupiennius salei, each particular predator-prey interaction could be related to obligatory behavioural patterns. The injection of venom by the spider was studied in relation to prey weight and time of injection. Single crickets (Acheta domesticus) of a given weight were offered successively to a spider at 10 min intervals. The prey items were removed 5, 60 or 180 sec after the beginning of the attack, and their state of activity recorded. Prey mortality was assumed to reflect the amount of venom injected. With relatively small prey items, the spider regulated this amount and venom glands were gradually emptied during successive bites. The injection was slow, lasting at least 3 min, possibly by a succession of discrete injections as a response to attempts by the prey to escape. In contrast, with larger prey items, the injection occurred within 1 min and mostly emptied the glands during the first bite. Bites experimentally reduced to 5 sec never allowed a spider to empty the venom glands completely. Only the first prey attacked usually received a lethal amount of venom. This suggested that an initial dose can be injected rapidly and could be morphologically segregated from the remaining venom.