The oviposition behavior of Tinthia myrmosaeformis Herrich-Schaeffer (Lepidoptera: Sesiidae), a candidate for the biological control of the invasive weed sulphur cinquefoil, Potentilla recta L., has been tested using three different test designs. Open-field, cage, and contact bioassays were carried out to test the host specificity of gravid females when they can display the complete sequence of pre- and post-alighting behavior, only the narrow-range pre-alighting and post-alighting phases, and exclusively the post-alighting phases. When gravid females were transferred by the experimenter onto test plants, all plant species offered were accepted to some extent. In the multiple-choice cage experiment, only a few test plant species outside the genus Potentilla were accepted; however, eggs were laid on the majority of test plant species within the genus Potentilla. The highest level of specificity was obtained in open-field experiments, where only the target weed and one other congeneric species were readily attacked. These results are largely in agreement with current models on host-selection behavior, and may be at least partly explained by different motivational thresholds experienced by the females in the different tests. A comparison of the three test designs suggests that the results from the bioassays in confinement were poor predictors of what T. myrmosaeformis accepted in the open-field experiment, but they point to a possible interaction between host fidelity and motivational state which may have remained undetected if only an open-field experiment had been carried out. (C) 2003 Elsevier Inc. All rights reserved.