Evaluators of aquatic systems often use benthic macroinvertebrates because they are central components in the food chain and useful for water-quality assessments and management decisions. More than 44 sampling devices are commonly used to collect benthic macroinvertebrates. Bitter Lake National Wildlife Refuge in New Mexico, USA, is a unique habitat with a complex of aquatic sinkholes. Morphometry, water chemistry, and biology differ vastly among sinkholes, creating challenges when deciding how to sample appropriately for benthic macroinvertebrates. We evaluated 5 sampling methods commonly used in lentic habitats: aquatic net (epibenthic sled), artificial substrate (Hester–Dendy multiplate [H-D] sampler), light trap, grab (Ponar grab), and suction sampler (pump sampler). We compared total number of taxa, % taxa detected, and total number of individuals collected among the 5 methods with a mixed-model incomplete block analysis of variance. The light trap caught the most taxa and yielded the highest % taxa detected. It also caught the most specimens. H-D samplers and Ponar grabs added taxa that were not commonly found in light traps, especially gastropods. We were unable to deploy the pump sampler and the epibenthic sled in all sinkholes, and they contributed little additional information. We recommend combined use of light traps, H-D samplers, and a Ponar grab to obtain a good representation of benthic invertebrates over a wide variety of environmental conditions in sinkholes and other lentic habitats. Together these methods detected 84 to 100% of taxa in each sinkhole.