This work advances the resilience-based management of small-scale fisheries facing both illegal fishing, climate change and cost uncertainties. It focuses on the coastal fishery of French Guiana in South America. Thus a dynamic, multi-species, resource-based and multi-fleet model including the illegal fishing is developed and calibrated using catch and effort time series over 2006-2018. Such model of intermediate complexity (MICE) also accounts for climate and energy costs stochasticities. From the calibrated model, fishing effort projections at the horizon 2050 are compared in terms of bio-economic resilience in the face of uncertainty scenarios. The bio-economic resilience metric is based on probabilistic viability (or reliability or robustness) involving different bio-economic thresholds related to biodiversity conservation, food security and profitability of fleets. It turns out that a massive reduction of the illegal fishing effort significantly improves this bio-economic resilience when compared to 'Business as Usual' (BAU) projections. However such a necessary enforcement approach against illegal fishing needs to be combined with a reallocation of the fishing efforts among the legal fleets to fully strengthen the bio-economic resilience of the whole fishery. Since the resilience-based management induces drastic changes, a 'transition' strategy accounting for the inertia of public policies and behavioral changes is also examined.