The variability in size structure and relative abundance (CPUE; number of fish ≥200 mm total length, L_T, collected per hour of electrofishing or trammel netting) of three native Colorado River fishes, the endangered humpback chub Gila cypha, flannelmouth sucker Catostomus latipinnus and bluehead sucker Catostomus discobolus, collected from electrofishing and trammel nets was assessed to determine which gear was most appropriate to detect trends in relative abundance of adult fishes. Coefficient of variation (CV) of CPUE ranged from 210 to 566 for electrofishing and 128 to 575 for trammel netting, depending on season, diel period and species. Mean CV was lowest for trammel nets for humpback chub (P = 0·004) and tended to be lower for flannelmouth sucker (P = 0·12), regardless of season or diel period. Only one bluehead sucker >200 mm was collected with electrofishing. Electrofishing and trammel netting CPUE were not related for humpback chub (r = −0·32, P = 0·43) or flannelmouth sucker (r = −0·27, P = 0·46) in samples from the same date, location and hour set. Electrofishing collected a higher proportion of smaller (<200 mm L_T) humpback chub (P < 0·001), flannelmouth suckers (P < 0·001) and bluehead suckers (P < 0·001) than trammel netting, suggesting that conclusions derived from one gear may not be the same as from the other gear. This is probably because these gears fished different habitats, which are occupied by different fish life stages. To detect a 25% change in CPUE at a power of 0·9, at least 473 trammel net sets or 1918 electrofishing samples would be needed in this 8 km reach. This unattainable amount of samples for both trammel netting and electrofishing indicates that detecting annual changes in CPUE may not be practical and analysis of long‐term data or stock assessment models using mark‐recapture methods may be needed to assess trends in abundance of Colorado River native fishes, and probably other rare fishes as well.