The objectives of this paper are twofold: first, to report our estimates of the meter‐to‐decameter‐scale topography and slopes of candidate landing sites for the Phoenix mission, based on analysis of Mars Global Surveyor (MGS) Mars Orbiter Camera (MOC) images with a typical pixel scale of 3 m and Mars Reconnaissance Orbiter (MRO) High Resolution Imaging Science Experiment (HiRISE) images at 0.3 m pixel⁻¹ and, second, to document in detail the geometric calibration, software, and procedures on which the photogrammetric analysis of HiRISE data is based. A combination of optical design modeling, laboratory observations, star images, and Mars images form the basis for software in the U.S. Geological Survey Integrated Software for Imagers and Spectrometers (ISIS) 3 system that corrects the images for a variety of distortions with single‐pixel or subpixel accuracy. Corrected images are analyzed in the commercial photogrammetric software SOCET SET (® BAE Systems), yielding digital topographic models (DTMs) with a grid spacing of 1 m (3–4 pixels) that require minimal interactive editing. Photoclinometry yields DTMs with single‐pixel grid spacing. Slopes from MOC and HiRISE are comparable throughout the latitude zone of interest and compare favorably with those where past missions have landed successfully; only the Mars Exploration Rover (MER) B site in Meridiani Planum is smoother. MOC results at multiple locations have root‐mean‐square (RMS) bidirectional slopes of 0.8–4.5° at baselines of 3–10 m. HiRISE stereopairs (one per final candidate site and one in the former site) yield 1.8–2.8° slopes at 1‐m baseline. Slopes at 1 m from photoclinometry are also in the range 2–3° after correction for image blur. Slopes exceeding the 16° Phoenix safety limit are extremely rare.