Introduction: Stereo imaging forms the basis for much of the 3-D terrain analysis conducted by researchers in the planetary science community. Identifying the data on which to conduct stereogrammetry can be complicated and time-consuming [1-3]. While some instrument teams maintain databases of deliberately targeted stereo-pairs (e.g., Mars Reconnaissance Orbiter (MRO) HiRISE [4] and Lunar Reconnaissance Orbiter (LRO) Camera (LROC) [5]) there is no tool to locate fortuitous stereo overlaps, especially for images from different instruments. Here we provide recommended methods and constraints for locating stereo pairs. Many of these recommendations can be tested using a new interactive solution provided by the USGS via the web-based Planetary Image Locator Tool (PILOT) application [6]. Overview: The main criteria to be used in identifying stereo images include: o Image overlap and similar spatial resolution. o 3-D stereo imaging “strength” as computed from emission and spacecraft azimuth angles. o Illumination similarity as computed from incidence and solar azimuth angles. o Similar solar longitude (i.e. it is best to avoid seasonal variations, such as differing frost patterns on Mars). o Compatible spectral wavelength range to achieve similar contrast. Recommendations are separated into two main categories of evaluation: (1) individually identify all candidate images that are suitable for stereo analysis and (2) identify images with common surface coverage that satisfy stereo pairing criteria. For both evaluation categories, the range of acceptable values depends on the intended use and thus most recommendations provided below are not absolute. While we provide specific “recommended” values, one should not take these as necessarily the “optimal” value because there can be a broad range of values that give similar quality without a sharp optimum in usefulness. Image Suitability for Stereo Analysis (1): Generally, criteria for finding useful stereo image candidates are generated for the center pixel of all possible images, although a more robust solution might compare only areas where images share common surface coverage. Incidence Angle. This angle is measured between the local surface normal vector at the surface intercept point (evaluated on a smooth representation of the global shape) and the vector to the Sun (for radar images, replace the sun vector with the radar source). o Limits: Between 40° and 65° depending on smoothness (shadows to be avoided) [7]. o Recommended: Nominally 50° Emission Angle. The angle is measured between the spacecraft-to-surface intercept vector and the local surface normal vector at the intercept point (evaluated on a smooth representation of the global shape). The goal of this criterion is to exclude images with extreme foreshortening (high emission angles for optical, low for radar). o Limits: Between 0° and the complement of the maximum slope (conservatively 45°, greater for smoother terrains) for optical images. Greater than the slope (≥15° even for smooth surfaces) for radar. o Recommended: No recommendation. Phase Angle (optional). Measured as the angle between the spacecraft-to-surface intercept vector and the illumination source (typically the Sun). Surface appearance can vary with phase angle, especially at low phase, so it may be useful to exclude low phase images. o Limits: Between 5° and 120°. o Recommended: ≥ 30° Ground Sampling Distance (GSD). The width of the pixels projected to the surface. o Limits: GSD is chosen based on the desired GSD of the output digital terrain model (DTM). Because typical stereo matching methods do not produce independent height estimates over distances smaller than about 3 to 5 image pixels, the image GSD needs to be 3 to 5 times smaller than the desired DTM GSD. o Recommended: Better than 1/3 of target DTM GSD.