The attraction of wild tephritids to semiochemical‐based lures is the ideal basis for trap network design in detection programmes, but in practice, mass‐reared colony insects are usually used to determine trap efficiency. For Bactrocera cucurbitae Coquillett, a lower response by wild males compared with colony‐derived individuals, usually used to estimate attraction parameters, could mean that the sensitivity of detection networks targeting this pest is reduced. We describe the results of mark–release–recapture experiments with wild‐ and colony‐derived B. cucurbitae males in a grid of cuelure‐baited traps within a macadamia nut orchard in Hawaii Island designed to quantify the attraction of cuelure to each fly type. For colony males, we estimate a 65% probability of capture at 27 m, intermediate with previous estimates on the attraction of methyl eugenol to Bactrocera dorsalis Hendel (36 m) and trimedlure to Ceratitis capitata Wiedemann (14 m) at the same site. Results suggest similar response over distance by wild‐derived B. cucurbitae compared with colony in the field, but there may be qualitative differences in response between wild and colony based on very low response of wild males in a standard bioassay of attraction. For both fly types, the estimates of attraction in the smaller of two grid sizes tested were lower than for the larger spacing, suggesting trap competition was a factor at an intertrap distance of 75 m. Dispersal patterns within the grid were generally to the south for the colony‐derived males and more variable for the wild‐derived males. In neither case was the direction of recapture correlated with the prevailing direction of the wind.