The New Caledonian crow is the only non-human animal known to craft hooked tools in the wild, but the ecological benefit of these relatively complex tools remains unknown. Here we show that crows acquire food several times faster when using hooked rather than non-hooked tools, regardless of tool material, prey type and extraction context. This implies that small changes to tool shape can strongly affect energy-intake rates, highlighting a powerful driver for technological advancement.New Caledonian (NC) crows Corvus moneduloides use a remarkable diversity of tool types for extractive foraging 1,2 , including 'non-hooked stick tools', which are simple unmodified twigs or leaf petioles 3 (Fig. 1a, bottom), and 'hooked stick tools', which are crafted from freshly-harvested branching twigs, often incorporating a range of distinct design features 1,[4][5][6][7][8][9] ( Fig. 1a, top; Supplementary Video 1). It has been suggested that hooked tool manufacture may have arisen through the (genetic or cultural) accumulation of innovations 6,9 (for a schematic illustration, see Fig. 1b). Any such cumulative process would require differential efficiency of tool types 10 -an important assumption we set out to test experimentally, addressing a conspicuous gap in our understanding of this avian model system 11 . To determine whether the use of hooks enhances food-intake rates of NC crows, we provided 17 temporarily-captive subjects with a suite of standardised, replicated extractive foraging tasks, and recorded the time taken to obtain embedded 'prey' when using either hooked or non-hooked stick tools (Fig. 1c; Supplementary Video 2). We compared time-toextraction data in a survival-analysis framework, with poor 'survival' of prey indicating a superior tool design. Crows manufactured their own hooked stick tools from fresh stems of the locally-preferred raw material, the perennial shrub Desmanthus virgatus 8 , and in a separate treatment, chose non-hooked stick tools from 100 assorted (non-forked) twigs and leaf petioles that had been collected from forest leaf litter. In each 90-minute trial, subjects were given wooden logs containing 18 extraction tasks, in which one of two prey types was hidden in one of two hole sizes (vermiform prey in narrow holes, vermiform prey in wide holes, and spiders in wide holes).Crow-made hooked stick tools (treatment 1a; solid blue lines in Fig. 1d) were three to thirteen times more efficient than crow-sourced non-hooked tools (treatment 1b; solid red lines), depending on the task (spiders in wide holes, HR = 2.91 [95% confidence interval = 1.62-5.24], z = 3.57, p < 0.001; vermiform prey in wide holes, , z = 5.64, p < 0.001; vermiform prey in narrow holes, .01], z = 6.34, p < 0.001). Although hooked tools were more efficient in all tasks, some tasks were inherently more demanding than others: spiders were acquired more quickly than vermiform prey from wide holes (for treatment 1a, HR = 1.79 [1.26-2.53], z = 3.25, p = 0.001), and vermiform prey more quickly from wide than from narrow holes (for t...