Brown bullhead catfish (Ameiurus nebulosus) were deliberately introduced to New Zealand from North America in the 1880s and populations have since increased significantly in the Auckland, Waikato, and Bay of Plenty regions. Catfish pose a major threat to New Zealand’s freshwater ecosystems, preying on and competing with native species for food, and decreasing water clarity through bioturbation and nutrient release. Early detection of the spread of catfish is crucial as it allows for containment and eradication measures to be set up, preventing the establishment of new catfish populations. However, traditional methods such as visual surveying, netting, and electrofishing are time-consuming and resource intensive, resulting in high costs and restricting the number of locations that can be surveyed. In addition, they can be relatively insensitive when the target biomasses are low, such as in newly establishing populations. False negatives can result in a lack of action from governmental agencies, allowing populations to establish and spread undetected. Domestic dogs (Canis familiaris) have previously been used for conservation management programmes, detecting reptiles, birds, insects, and larger wildlife animals. Although the use of dogs to detect targets in aquatic environments is relatively new, research has indicated domestic dogs have the ability to discriminate between similar aquatic species and previous projects related to catfish scent detection have shown domestic dogs can detect the presence of catfish in dechlorinated water samples at biologically relevant concentrations. The aim of this research was to determine whether domestic dogs could detect the presence of catfish in lake water samples at a biomass concentration equivalent to 43.5 kg/ha (based on the assumption of a 2 m deep waterbody), a concentration consistent with the estimated populations of catfish in New Zealand lakes. The first stage of the experiment determined whether dogs could detect the presence of catfish in dechlorinated water samples at a biomass concentration of 43.5 kg/ha. Water samples were collected from tanks containing a standard catfish biomass concentration of 15.5 g/L (equivalent to 38,700 kg/ha) and diluted to 311 kg/ha. As the dogs achieved the discrimination criteria, the sample concentration was progressively decreased. All three dogs were able to detect the presence of catfish at a biomass concentration of 43.5 kg/ha. The next stage of the experiment determined whether domestic dogs could detect the presence of catfish in lake water samples. Lake water samples were collected from catfish-absent Lake Rotomā and spiked with catfish aquaria. As the dogs achieved the discrimination criteria, the catfish concentration was progressively diluted. The majority of the dogs in this study successfully detected the presence of catfish at a biomass concentration of 43.5 kg/ha in lake water samples, with one dog successfully detecting catfish at a biomass concentration of 1.55 kg/ha. In New Zealand, it is highly unlikely significant environmental impacts would occur from a catfish biomass of 1.55 kg/ha. However, targeted eradication of a population density this size would be highly advisable and may be feasibly managed depending on available resources, size of the target water body and its connections to other waterways. The findings in this study indicate the potential utility of dogs in the early detection and management of invasive freshwater species. According to the theory of generalisation, the performance of these dogs may be generalised to different lakes and to different biomass equivalent concentrations without specific training on those lakes or concentrations. However, it would be important to test for this generalisation in future research as differences in stimuli can affect generalisation, with less generalisation occurring as stimuli become increasingly dissimilar from the original training stimulus. Therefore, the differences between the conditions in this study and those in an operational scenario will be important in determining whether the obtained performance will generalise to the operational scenario.