Pups on San Nicolas Island were counted by two methods; counts by observers on the ground were compared to counts from aerial photographs taken with a 126-mm-format camera with image motion compensation. No difference was detected between photographic counts and ground counts (P= 0.367) when ground counters had unobstructed views. However, ground counts were significantly lower when areas with obstructed views were included in the analysis (P < 0.001). For areas with unobstructed viewing conditions, no difference was detected between counts by the two methods for rock substrates (P= 0.140), sand substrates (P= 0.468), or mixed rock-and-sand substrates (P= 0.968). No differences were found among three replicate aerial photographic censuses (P= 0.432), but a significant difference was found between two replicate ground censuses (P= 0.037). Total counts obtained from the aerial photographs were more precise (CV = 0.042) than counts obtained on the ground (CV = 0.078). Less variability in counts was found between photographic counters than for ground counters.