Midwater fish faunas over three cold seep habitats >1000m (AC601, GC852, AT340) and over a cold-water coral bank (VK826, <1000m) in the north-central Gulf of Mexico (GOM) are described and compared. Day and night discrete-depth sampling was accomplished at each study site, 9–29 August 2007, using an opening–closing Tucker trawl towed at depth for 30min. The 159 Tucker trawl stations yielded 9802 individuals and at least 126 species (30 families) of juvenile and adult midwater fishes. Myctophidae (38 species, 28% of the total catch), Stomiidae (17 species, 1.4%), Gonostomatidae (12 species, 56%), Sternoptychidae (10 species, 6%), and Phosichthyidae (5 species, 6%) dominated catches at all sites. Thirty-one species (8 families) comprised about 90% of the overall abundance of fishes across the four sites. Despite the wide separation of study sites and some differences in physical oceanography, geographic and temporal patterns in mesopelagic fish assemblages were not evident (ANOSIM, Global R=0.20, p=0.1%, pairwise R values <0.3). However, station groupings were related mostly to similarities in sampled depths (SIMPROF, p<1.5%). Upper water-column station groups (<400m) were typified by Lepidophanes guentheri, Diaphus dumerilii, Vinciguerria nimbaria, Vinciguerria poweriae, Valenciennellus tripunctulatus, Argyropelecus aculeatus, Notolychnus valdiviae, Benthosema suborbitale, Gonostoma elongatum, and Hygophum taaningi. Deeper station groups (mean depth=609m) were typified by Cyclothone alba, Cyclothone acclinidens, Cyclothone braueri, Cyclothone obscura, Cyclothone pallida, Cyclothone pseudopallida, V. tripunctulatus, and Hygophum benoiti. Depth, diel, and size distributions were also analyzed for abundant species. For example, Cyclothone spp., abundant at all sites except VK826, usually occurred in deeper waters (>130m), especially during daylight hours (>335m). Most Cyclothone individuals were captured at night at GC852, whereas catches were more evenly distributed across times at AT340 and VK826. Cyclothone spp. individuals occurred in the deeper parts of the sampled depth range at AT340 and VK826 at all times compared with GC852, where they appeared to undergo a diel migration. The two most abundant Cyclothone spp., C. pallida and C. pseudopallida, had the broadest depth ranges (150–1377m) and exhibited a trend for larger specimens to occur deeper at GC852. Most V. tripunctulatus were captured at night and tended to have larger mean sizes in deeper waters. This species did not exhibit strong diel patterns of vertical distribution. Shallower depths were occupied at night by some individuals at all sites, but most individuals remained somewhat deeper at night at VK826 (350–450m at night and 200–350m by day) and AT340 (200–600m at night and 300–500m by day) compared with GC852 (150–450m at night and 750 and 1000m by day). Nearly all Vinciguerria poweriae were caught at night, and no clear trend of size with depth was apparent. This species occupied a wider depth range at night, with most in the upper 200m. Most myctophids, the strongest diel migrators, were captured at night at the offshore stations. The dominant myctophid, N. valdiviae, was most abundant at GC852 (550–1050m by day and 50–750m at night), but diel movement was more distinct at AT340 (<200m at night and 350–500m by day). No size-related trends with depth were apparent at any site for N. valdiviae, and size distributions were similar among sites.This study and others support the hypothesis that there is a geographically similar mesopelagic fish fauna throughout the GOM. However, the fauna does vary by depth, with some species remaining in deep waters at all times (e.g., Cyclothone spp.) and others occurring at mid-depths and exhibiting varying degrees of diel migration (e.g., many myctophids). Even though faunal composition is similar across the GOM, some species exhibit different movement behaviors at different locations. Such variances could be related to differences in physical or chemical factors, food resources, lunar phase, or bottom depth.