Abstract. 1. Repeat surveys are needed to capture a representative spectrum of adult odonate richness at a site, but specifics on frequency and duration of surveys and associated inferential biases are poorly understood.
2. Weekly 1 h surveys of mature male dragonflies and damselflies were repeated at least 15 times at 19 ponds, lakes and wetlands scattered throughout North America. For each site, we tallied the data remaining when the weekly frequency was reduced to 75% (every 1.5 weeks), 50% (biweekly), 33% (triweekly), and 25% (monthly) and the 1 h survey to 50, 40, 30, 20 and 10 min subsets.
3. Reducing the original effort by half (i.e. to 30 min biweekly) retained about 80% of the species on average. The smallest effort (10 min monthly) retained about 49% of species. The greatest rate of information loss occurred between 20 and 10 min.
4. Across-site analysis found that data subsets correlated to the original data set (r > 0.81) despite up to 50% species loss. Strong correlations (r ≥ 0.98) remained with 10–15% species loss.
5. Biweekly surveys lasting 20–40 min each may provide a representative and cost-effective sample of adult odonate richness in lentic study sites. Losing a handful of species should not greatly undermine richness and compositional comparisons among sites.