Efficiency and reliability of a suction device for assessingRhinocyllus conicuspopulations were determined and compared with current visual research techniques. A suction machine collected a greater proportion of weevils from thistles ranging in height from 6 cm to 1 m but was less efficient than visual searches on rosettes or plants in excess of 1 m. A single, 2-sec suction sample was as efficient as multiple samples with durations up to 6 sec. Horizontal approaches were superior to up-sweep or vertically oriented suction samples and collected 98.6 ± 1.3% of adult weevils in recapture tests.R. conicuspopulation density increased with plant height; correlation coefficients were highest (r = 0.95) when plant height variability was greatest and all thistles except rosettes were sampled. Although ca. 15% of the potentially viable eggs per primary bloom may be dislodged during suction sampling, removal of old, dried-out egg casings resulted in counts of eggs remaining on thistle blooms which were generally more indicative of actual weevil populations.