Abstract: During April and May 2007, a rodenticide experiment was conducted on control of plateau pika in Sunan, Gansu province. Recommended drug densities and drug values of five rodenticides were adjusted by geometrically increased and decreased values using an orthogonal design. There were differences between recommended values, actual drug densities, drug consumption, bait types, and of adjusted values and types in killing efficiency on plateau pika in the test region. Primary and secondary factors affected killing efficiency and tended to decrease in the order rodenticide types＞drug densities＞bait consumption＞bait types. Five kinds of tested rodenticides were in the order Talon (83.9%)＞Botulin type D (82.6%)＞Bromadiolone (79.4%)＞Botulin type C (75.0%)＞Diphacinone sodium salt (68.9%) according to the assessment index based on control effect. The three optimum treatment combinations were 0.15% Botulin type D, wheat bait, 25 grains bait; 0.01% Talon, naked barley bait, 5 g bait; 0.01% Bromadiolone, wheat bait, 3 g bait.