Abstract A trial was conducted to assess the repellent properties of three materials and of crude hexane and methanol extracts of the fruit of the osage orange (extracts were tested individually and in combination). Pyrethrins were used as a standard for comparison. Topical toxicity of the osage orange extracts was also examined by applying 0.1, 1.0, 10, or 100 μg of the extracts, dissolved in acetone, to the abdominal venters of adult male cockroaches. The extracts were prepared by shaking 200 g masticated osage orange fruit with 200 ml hexane or methanol for 30 min. The extract and solvent were then filtered through Na2S04. Shaking and filtering were repeated three times. The solvent was evaporated from the resultant extracts via rotary evaporation. To evaluate repellency, a choice-box method was used. The choice box consisted of two clear, plastic chambers (9 × 8.5 × 2 cm), connected via 20 cm of 1-cm-diameter Tygon® tubing. Five adult male cockroaches (2 wk post-molt, wild strain maintained in the laboratory for 5 yr) were placed in each of the two chambers of each choice box for 24 h prior to the start of the experiment. Filter paper (5 cm, Whatman no. 5) bearing small fragments (< 0.5 cm2) of the materials, or treated with the extracts or pyrethrins, was then placed in one randomly chosen chamber of the choice box and untreated filter paper was placed in the other. Treatment concentrations were expressed as parts per million (ppm) based on μg of material added per unit (cm3) of the treated chamber. Treatments were replicated six (for bay leaves, chewing gum, osage orange fruit, and pyrethrins) or seven (for the hexane and methanol extracts) times. Locations of the cockroaches with respect to the treated chambers were then noted at 1, 4, 24, and 48 h following exposure. No food or water was supplied during the experiment and choice boxes were held under 12:12 L:D. Data were expressed as mean % repelled from treated chamber; these data were transformed prior to analysis to arcsin %. Analysis of variance (ANOVA) and Duncan's multiple range test were used to detect and delineate treatment differences.