Ground observations of 6300 ¿ auroral emission at the polar cap boundary are studied to determine the accuracy with which the latitudinal profile of emission intensity can be used to identify the separatrix. Meridian scanning photometers at Rankin Inlet and Gillam provide the observations of 6300 ¿ emission, and the separatrix determinations obtained from the photometer data are compared with those determined from DMSP F9 precipitating particle data obtained within 1.5 hours in MLT of the ground stations. We assume that the separatrix lies at the poleward edge of boundary plasma sheet precipitation. We find that the average intensity of 6300 ¿ in the polar cap is fairly uniform at 60 R. In the auroral zone the average emission is fairly uniform at 170 R. On the basis of the efficiency and accuracy of separatrix identification, choosing a threshold of 110 R provides the best identification of the polar cap boundary. The rms error in this identification is 1.2¿, and the boundary is located in 54% of the cases. The latitudinal gradient of the emission intensity is also investigated as a possible identifier of the polar cap boundary. However, using the intensity gradient to identify the polar cap boundary is less accurate, with a minimum uncertainty of 2.4¿. Finally, fitting the measurements of the 6300 ¿ emission to a latitudinal step function, which represents an idealized emission profile, reduces the rms error in the identification of the separatrix to 1.0¿ while still identifying the boundary in 54% of the cases.¿ 1997 American Geophysical Union