The native freshwater banded kokopu (Galaxias fasciatus Gray) is generally regarded as a nocturnal predator of small native forest streams. Current methods for determining absolute abundances of this species in these streams is usually by the expensive, labour intensive, and invasive method of electric fishing. Spotlighting at night has previously been used to determine their presence and relative abundance, although the proportion seen (the bias of the method) was unknown. This bias was quantified by determining the numbers of banded kokopu seen at night in 20 m reaches against population estimates made in the same reaches the following day by removal electric fishing. This showed that, on average, only around 64% of fish of all sizes were being seen. Counts made by spotlighting therefore should be multiplied by 1.57 to correct for this bias and to estimate for the actual number of fish present. Spotlighting was found to be less efficient for smaller fish (approximately <60 mm total length), with only around 42% being observed. Spotlight counts for smaller fish therefore needed to be multiplied by 2.41 to estimate the actual abundances of fish of this size. Fish larger than this size, however, were more often seen (0.70) requiring their spotlight counts only to be multiplied by 1.42 to correct for their spotlight estimates. Banded kokopu diel activity was studied using time-lapse video recordings with a low-light sensitive camera placed above pools in natural streams. Two major activity peaks were noted; one from 6:00 a.m. to around 8:00 a.m., the other from around 2:00 p.m. to around 7:30 p.m. Foraging behaviour, intra- and interspecific behaviours, and mode of swimming were all also noted. Nocturnal habitat preferences and suitability functions were derived for all sizes of banded kokopu together, and then for the two aforementioned size classes separately. The following variables were tested for; mean column water velocity, surface water velocity, column depth, substrate size, and cover type. Used habitat data was collected at night by a colour filtered spotlight, with a small floating marker being placed at the focal point of the fish. The particular characteristics of this point were then measured along with available habitat data the following day. Fish were found to have a very narrow preference for very slow water velocities (around 0.05 m/s), with juveniles preferring slightly faster than adults when considered separately. Preferred water depths for both sizes were around 0.8 m, and preferred substrate sizes were fine gravels, sand, and silt. Preferred cover for all sizes was for tree and tree fern roots and undercut banks, as it also was for larger fish when considered separately. Smaller fish did not use tree fern roots significantly, but instead preferred the cover of cobbles. This smaller sized fish also made use of tree roots and undercut banks. These new banded kokopu habitat preference data were then applied to an IFIM situation in the Waitakere River (West Auckland), where Resource Consent for abstraction of 0.0041 m³/ s (total daily abstraction of 119 m³) was being requested for irrigation of the adjacent Waitakere Golf Course. The headwaters of this river are situated in the pristine native forest of the Waitakere Ranges. This river also contains a significant dam (the Waitakere Reservoir) a few kilometres upstream of the proposed site of abstraction which diverts 16, 000 m³ a day to out of stream uses. This figure represents approximately 80% of water that would otherwise continue downstream and eventually to the ecologically significant freshwater Te Henga Wetland. Changes in weighted useable area (WUA) as flow decreased as a result of this abstraction were modelled on the DOS program “RHYHABSIM” for fish and invertebrates known to be in this river. A comparison was also made of this habitat flow method with the North American Tennant method; an historical flow method which determines the effects of flows by set proportions of the mean low flow. No substantial reduction in WUA was noted for any of the species following abstraction from either a mean low flow (0.0200 m³/s), or a one-in-five-year low flow (0.0134 m³/s). Endorsement was therefore given for the proposed abstraction to Receive Consent. Recommendations on the time of day that the abstraction should take place, and other further methods to mitigate the ecological effects of this abstraction are also made.