Dietary mixing, which means that polyphagous animals prefer to switch the diet items during ontogeny, often affects their growth and developmental performance positively. Two hypotheses, known as nutritional compensation and toxin dilution effect, have been put forward to explain such positive effect based upon the observations of various grasshopper species. As have been demonstrated in many Lepidopterous species, however, polyphagy at the species or population levels do not necessarily imply polyphagy at the individual level, for a caterpillar often has no chance to choose among host plant species due to temporal and spatial separation. In this paper, the authors reported the larvae feeding behavior of a typical generalist species, Helicoverpa armigera (Hübner), by means of forced host-switching (at two hour interval) and cross-coating of secondary metabolites (at natural concentrations) from four common host plant species of H. armigera larvae: tobacco, hot pepper, tomato, and cotton. The results show that, in all the cases, the dietary-mixing did not significantly increase the overall leaf consumption of the fourth instar larvae, moreover, some combinations inhibited feeding amounts and others had no effect, suggesting that nutritional compensation hypothesis could not explain the dietary mixing behavior of H. armigera larvae. Both tobacco leaf and its major secondary metabolite, nicotine hydrate, significantly inhibited the leaf consumption of all the other host species, while neither hot pepper leaf nor its secondary metabolite, (E)-capsaicin, showed effect on feeding amounts of all the other host species. Although tomato leaf in dietary-switching groups did not affect the consumptions of all the other three host leaves, α-tomatine showed significant feeding deterrence when coated on the leaf of hot pepper or cotton. The consumption of tobacco leaf discs increased in cotton-tobacco switching group compared with that in tobacco non-switching group, but the coating of (+)-gossypol on tobacco leaf discs decreased the feeding amounts. Cotton and hot pepper leaf showed no significant interactive effect, neither in force-switching bioassay nor in cross-coating bioassay. Taken together, H. armigera larvae may cope with the secondary metabolites contained in their natural host plants by tolerance strategy, and the tolerance level of an individual larva is in accordance with the symmorphosis hypothesis, i.e. a certain individual larva does not necessarily evolve the mechanisms to adapt multiple secondary metabolites which derived from all the recorded host species.