Individuals exhibiting a high level of cognitive ability may also exhibit more elaborate traits and so gain higher levels of mating success. This suggests that selection may act on cognitive performance through mate choice. Studies investigating this relationship have tended to focus on single cognitive tasks, or tasks that are closely related to existing natural behaviours, and individuals are frequently tested in captive conditions. This can introduce test artefacts and may tell us more about selection on specific display behaviours that we imagine being particularly cognitively complex, rather than a general cognitive ability. We tested free-living male spotted bowerbirds, Ptilonorhynchus maculatus, that exhibit elaborate sexual displays which appear to be cognitively demanding. We describe a method for testing individuals in the wild, without the need for constraint or captivity. We looked for evidence of a general cognitive ability in males by assaying their performance in a series of novel tasks reflecting their natural bower-building behaviour (bower maintenance) or capturing more abstract measures of cognitive ability (colour and shape discrimination, reversal learning, spatial memory and motor skills). We related performance in these tasks to their mating success. An individual's performance in one task was a relatively poor predictor of performance in any other task. However, an individual's performance across tasks could be summarized by a principal component which explained a level of total variance above which has previously been accepted as evidence of a general cognitive ability. We found no relationships between an individual's overall performance, or performance in any single task, and mating success. Our results highlight the need for further investigation of whether selection on cognition in bowerbirds is exerted through mate choice. We offer this as an example of how classic cognitive tasks can be transferred to the wild, thus overcoming some limitations of captive cognitive testing.