Attractive targeted sugar bait (ATSB) is a potential new vector control tool that exploits the sugar-feeding behaviour of mosquitoes. Little is known about the factors which drive ATSB efficacy, either as a standalone vector control tool or in combination with existing intervention strategies. It has been suggested that the percentage of wild mosquitoes caught fed on dye-containing sugar baits without the toxin could provide an entomological correlate of the potential epidemiological benefit of ATSB. A transmission dynamics mathematical model is combined with data from wild mosquitoes to investigate the relationship between the mosquito dyed fraction, bait-feeding rate and the potential epidemiological impact of ATSB in the presence of standard malaria control. The dyed fraction in Mali varies substantially in space and time (mean 0.34, standard deviation 0.15), causing estimates of the bait-feeding rate to be highly uncertain, especially in areas with existing vector control tools. The model indicates the dyed fractions observed in field experiments were broadly predictive of the reductions in mosquitoes caught when ATSB stations were deployed at scale in Mali (R 2 = 0.90). Model projections suggest that if these bait-feeding rates were observed in all mosquitoes, then the widespread use of ATSB could substantially reduce malaria burden alone or in combinations with standard malaria control, though epidemiological impact is likely to vary substantially in different areas. For example, observing a dyed fraction of 5% would indicate a daily bait-feeding rate of 0.024 (range 0.008-0.049) which is projected to result in 0.13 clinical cases averted per person-year (range 0.051-0.22), a 39% efficacy (range 12-66%) in this particular site. Nevertheless, the uncertainty in the relationship between the observed dyed fraction and the true bait-feeding rate, and the underlying biology of mosquito sugar-feeding means that the epidemiological benefit of this new possible intervention remains unclear.