Despite much recent interest, it is unclear which types of landmarks are best suited for survey knowledge acquisition. Thus, we investigated the accuracy of survey knowledge after the learning of sequentially visible (local) landmarks and simultaneously visible (global) landmarks from a first-person perspective during navigation through a virtual city. We also assessed systematically the role of working memory during navigation with a concurrent spatial-sequential task. Our results indicate that the learning of local and global landmark configurations are similarly affected by this concurrent task. We also find greater accuracy for the acquisition of global landmark knowledge compared to local landmark knowledge, especially for participants with high working memory capacity.