We tested three hypotheses concerning the function of stabilimenta in the orb weaver Gasteracantha cancriformis: 1) warning to large animals that might accidentally destroy the web; 2) prey attraction; and 3) camouflage. One prediction of the warning hypothesis (but not of the others), that less exposed webs should have fewer stabilimentum tufts, was verified: very few tufts occurred on the peripheral lines of small webs. On the other hand, a prediction of the prey attraction hypothesis, that webs with more stabilimentum tufts should also have more captured prey, was only confirmed in one sub sample, and further analysis indicated that spider size rather than number of stabilimentum tufts best explained the presence of prey. An additional observation not in accord with prey attraction was that resting webs, which lacked sticky silk for prey capture, nevertheless had abundant stabilimentum tufts. Finally, the number of stabilimentum tufts was lower in the webs of white (as opposed to yellow or orange) spiders, contradicting a prediction of the camouflage hypothesis. The strongest conclusions from our results are support for the warning function, and lack of support for the prey attraction function.