Not all conservation interventions are successful at correcting threatening processes and the odds of failure increase with uncertainty concerning the true threats to a population. Failure of conservation actions to improve demographic rates might be evidence of their ineffectiveness, or that other unaddressed threats nullify the potential benefits of interventions. Knowledge of key threatening processes that afflict Orange-bellied Parrots Neophema chrysogaster is lacking, but population modelling predicts that actions in the breeding range are unlikely to correct decline unless mortality during migration/wintering is addressed. Despite this, there has been a spatial bias in recovery effort towards the breeding range in recent decades. We model annual survival data spanning 1995–2017 for the last known wild population to evaluate whether the predictions about the efficacy of recovery efforts are accurate. Based on our best-supported model, probability of adult survival was constant at 0.58, but juvenile survival declined from 0.51 to 0.20. Survival did not improve when we considered the effects of recovery actions in the breeding grounds (which only aimed to correct local scale threats anyway). This result supports predictions that conservation interventions in the breeding ground alone are not sufficient to recover this species. We conclude that although interventions in the breeding ground may have corrected local threats, birds succumbed to other threats during migration/winter. It is crucial that new targeted interventions be identified and implemented to reduce mortality of Orange-bellied Parrots in their migration/winter habitats to prevent extinction.