During a volcanic crisis, evacuation is the most effective mitigation measure to preserve life. However, the decision to call an evacuation is typically complex and challenging, in part due to uncertainties related to the behaviour of the volcano. Cost-benefit analysis (CBA) can support decision-makers: this approach compares the cost of evacuating versus the expected loss from not evacuating, expressed as a ‘break-even’ probability of fatality. Here we combine CBA with a Bayesian Event Tree for Short-term Volcanic Hazard (BET_VHst) to create an evacuation decision-support tool to identify locations that are cost-beneficial to evacuate in the event of volcanic unrest within a distributed volcanic field. We test this approach with the monogenetic Auckland Volcanic Field (AVF), situated beneath the city of Auckland, New Zealand. We develop a BET_VHst for the AVF, extending a recently revised Bayesian Event Tree for Eruption Forecasting (BET_EF) to consider the eruptive style, phenomena produced, and the impact exceedance probability as a function of distance. The output of the BET_VHst is a probability of volcanic hazard impact at a given location. Furthermore, we propose amending the weight of the monitoring component within the BET_VHst framework to a transitional parameter, addressing limitations identified in a previous study. We examine how three possible transitional monitoring component weights affect the spatial vent likelihood and subsequent BET_VHst outputs, compared to the current default weight. For the CBA, we investigate four thresholds, based on two evacuation durations and two different estimates for the value of life that determine the cost of not evacuating. The combinations of CBA and BET_VHst are tested using a synthetic unrest dataset to define an evacuation area for each day. While suitable evacuation areas were identified, there are further considerations required before such an approach can be applied operationally to support crisis management.