The Mars Descent Imager, or MARDI, experiment on the Mars Polar Lander (MPL) consists of a camera characterized by small physical size and mass (~6 ¿ 6 ¿ 12 cm, including baffle; <500 gm), low power requirements (<2.5 W, including power supply losses), and high science performance (1000 ¿ 1000 pixel, low noise). The intent of the investigation is to acquire nested images over a range of resolutions, from 8 m/pixel to better than 1 cm/pixel, during the roughly 2 min it takes the MPL to descend from 8 km to the surface under parachute and rocket-powered deceleration. Observational goals will include studies of (1) surface morphology (e.g., nature and distribution of landforms indicating past and present environmental processes); (2) local and regional geography (e.g., context for other lander instruments: precise location, detailed local relief); and (3) relationships to features seen in orbiter data. To accomplish these goals, MARDI will collect three types of images. Four small images (256 ¿ 256 pixels) will be acquired on 0.5 s centers beginning 0.3 s before MPL's heatshield is jettisoned. Sixteen full-frame images (1024 ¿ 1024, circularly edited) will be acquired on 5.3 s centers thereafter. Just after backshell jettison but prior to the start of powered descent, a best final nonpowered descent image will be acquired. Five seconds after the start of powered descent, the camera will begin acquiring images on 4 s centers. Storage for as many as ten 800 ¿ 800 pixel images is available during terminal descent. A number of spacecraft factors are likely to impact the quality of MARDI images, including substantial motion blur resulting from large rates of attitude variation during parachute descent and substantial rocket-engine-induced vibration during powered descent. In addition, the mounting location of the camera places the exhaust plume of the hydrazine engines prominently in the field of view. ¿ 2001 The American Geophysical Union.