The thicknesses obtained from drill-hole measurements on four very thick multi-year ice floes are compared to thicknesses obtained over the same profile areas by a helicopter-based electromagnetic induction (HEM) system. Drill-hole measurements produced average thicknesses of 8.3 m, 6.5 m, 10.2 m and 12.9 m. Compared to individual drill-hole measurements, the HEM overestimated the percentage of ice from 3 to 7 m thick, and did not reproduce thicknesses larger than about 12 m. Since important information about the maximum thickness of the most massive ice features was not captured, it caused the average thickness of very thick multi-year floes to be underestimated by the HEM. The average thickness of the sampling areas on the four examined floes was underestimated by 1.2 to 3.1 m, or from 15 to 24%, with the thickest, most deformed ice floe producing the least favourable agreement. The paper shows that the HEM provides a reasonable estimate of the average thickness of deformed multi-year ice when the ice is less than about 10 m thick, on average however, it should also be noted that 5 of the 24 multi-year floes (20%) on which more than 600 drill-hole measurements have been made over the past three years have had an average thickness of 10 m, or more. Evidence suggests that HEM surveys may be missing a component of Arctic sea ice that is important for offshore operations and seafloor scouring.