Conservation and management actions are often highly dependent on accurate estimations of population sizes. However, these estimates are difficult to obtain for elusive and rare species. We compared two census methods for Eurasian otter: snow tracking and noninvasive genetic census based on the genotyping of faecal samples. With the noninvasive genetic census we detected the presence of almost twice as many otters as with snow tracking (23 and 10-15, respectively), and mark-recapture estimates based on the genetic census indicated that the real number of otters could be even higher. Our results indicate that snow tracking tends to underestimate the number of individuals and also that it is more susceptible to subjective assessment. We compared the strengths and weaknesses of the two methods.