The recent emergence of wide-bandgap power transistors enables higher switching frequencies in electrical motor drives. Their full utilization from a system point of view requires quantification of the corresponding time-harmonic motor losses. As an initial step, this paper presents a unique study of stator losses for three different commercially available nonoriented silicon–iron steel grades (with lamination thicknesses 0.1, 0.2, and 0.3 mm). The investigations cover a wide frequency range (10–100 kHz) at different levels of dc bias (up to 1.6 T). Iron losses are identified from measurements on fully assembled stators deploying a novel technique. By utilizing fully assembled stators, no additional samples are required. Manufacturing influence is inherently incorporated. Results show that measured iron losses are twice as high at 10 kHz compared with Epstein test results, which emphasizes the need to incorporate manufacturing influence on iron losses at high frequencies. The level of dc bias is also observed to have a significant impact on iron losses (up to 30%). Even though thinner laminations are known for reducing iron losses, the reduction is much lower than anticipated in the studied frequency range due to skin effect. Using a 0.1-mm lamination gauge instead of 0.3 mm reduces losses by 50% at 10 kHz, while the same substitution at 100 kHz only reduces losses by 30%. Future work includes loss separation in complete converter-fed machines.