A terrain‐aided navigation method for an underwater glider is proposed that is suitable for use in ice‐covered regions or areas with heavy ship traffic where the glider may not be able to surface for GPS location updates. The algorithm is based on a jittered bootstrap algorithm, which is a type of particle filter that makes use of the vehicle's dead‐reckoned navigation solution, onboard altimeter, and a local digital elevation model (DEM). An evaluation is performed through postprocessing offline location estimates from field trials that took place in Holyrood Arm, Newfoundland, overlapping a previously collected DEM. During the postprocessing of these trials, the number of particles, jittering variance, and DEM grid cell size were varied, showing that convergence is maintained for 1,000 particles, a jittering variance of 15 , and a range of DEM grid cell sizes from the base size of 2 m up to 100 m. Using nominal values, the algorithm is shown to maintain bounded error location estimates with root‐mean‐square (RMS) errors of 33 and 50 m in two sets of trials. These errors are contrasted with dead‐reckoned errors of 900 m and 5.5 km in those same trials. Online open‐loop field trials were performed for which RMS errors of 76 and 32 m‐ were obtained during 2‐h‐long trials. The dead‐reckoned error for these same trials was 190 and 90 m, respectively. The online open‐loop trials validate the filter despite the large dead‐reckoned errors, single‐beam altitude measurements, and short test duration.