This study evaluates the technical efficiency of saffron farms in Torbat Heydarieh and Zaveh Counties, ‎located on northeastern Iran, using a two-stage double-bootstrap Data Envelopment Analysis (DEA) ‎approach. Cross-sectional data were collected from 108 saffron producers across six districts, and ‎technical efficiency (TE) scores were estimated under variable returns to scale (VRS) with an input-‎oriented DEA model. The bias-corrected double-bootstrap procedure of Simar and Wilson (2007, ‎Algorithm 2) was applied with 1000 replications to ensure consistent inference and robust estimation. ‎Results reveal that the Farrell efficiency estimates tend to overstate farm technical efficiency by ‎approximately 25 percent. I detected significant regional disparities in efficiency scores (χ² = 12.74, p < ‎‎0.05), with the highest mean efficiency observed in Kadkan (0.32) and the lowest in the Central district ‎‎(0.17). The truncated bootstrap regression identified an inverted U-shaped relationship between farm ‎age and efficiency, where efficiency initially increases with farm age but declines beyond a certain ‎point. Experience in saffron cultivation had a strong positive and statistically significant effect on ‎efficiency (p < 0.01), while education level and farmer’s age were not significant predictors. Slack ‎analysis revealed considerable input-specific inefficiencies and spatial heterogeneity in slacks. On ‎average, farms could reduce water use by 706.88 m³/ha, land by 3.04 ha, corm use by ‎528.10‎ kg/ha, and ‎labor by 20 person-days/ha without reducing output.‎