Experimental work increasingly suggests that non-random pathogen associations can affect the spread or severity of disease. Yet due to difficulties distinguishing and interpreting co-infections, evidence for the presence and directionality of pathogen co-occurrences in wildlife is rudimentary. We provide empirical evidence for pathogen co-occurrences by analysing infection matrices for avian malaria (Haemoproteus and Plasmodium spp.) and parasitic filarial nematodes (microfilariae) in wild birds (New Caledonian Zosterops spp.). Using visual and genus-specific molecular parasite screening, we identified high levels of co-infections that would have been missed using PCR alone. Avian malaria lineages were assigned to species level using morphological descriptions. We estimated parasite co-occurrence probabilities, while accounting for environmental predictors, in a hierarchical multivariate logistic regression. Co-infections occurred in 36% of infected birds. We identified both positively and negatively correlated parasite co-occurrence probabilities when accounting for host, habitat and island effects. Two of three pairwise avian malaria co-occurrences were strongly negative, despite each malaria parasite occurring across all islands and habitats. Birds with microfilariae had elevated heterophil to lymphocyte ratios and were all co-infected with avian malaria, consistent with evidence that host immune modulation by parasitic nematodes facilitates malaria co-infections. Importantly, co-occurrence patterns with microfilariae varied in direction among avian malaria species; two malaria parasites correlated positively but a third correlated negatively with microfilariae. We show that wildlife co-infections are frequent, possibly affecting infection rates through competition or facilitation. We argue that combining multiple diagnostic screening methods with multivariate logistic regression offers a platform to disentangle impacts of environmental factors and parasite co-occurrences on wildlife disease.