This study is aiming to propose a framework integrating new criteria system, rough-fuzzy best-worst method (BWM) and rough-fuzzy data envelopment analysis (DEA) for evaluation and selection of sustainable hydrogen production technologies (HPTs) with hybrid types of sustainability indicators (i.e. quantitative and qualitative) and hybrid decision uncertainties (i.e. intrapersonal linguistic vagueness and interpersonal preference randomness). The rough-fuzzy BWM is presented to determine the relative weights of sustainability criteria and the rough-fuzzy DEA is proposed for prioritizing the alternative HPTs with hybrid types of performance data (including crisp number, interval number and group linguistic terms). Such integration combines the collaborative capability of rough-fuzzy number in fully manipulating hybrid uncertainties, the merits of BWM in quickly searching optimal weights and the advantages of DEA in effectively obtaining efficiency with multiscale input and output data. The application of the proposed approach to five typical HPTs with 14 hybrid uncertain sustainability indicators show that the wind-based electrolysis has the most sustainability priority (0.480) for hydrogen production compared with coal gasification (0.035), steam methane reforming (0.056), biomass gasification (0.246) and photovoltaic-based electrolysis (0.182). Moreover, the comparisons with other methods demonstrate the validity and effectiveness of the proposed approach.