The article summarizes results of part of the second VAMAS a.c. loss measurement intercomparison. This program was carried out at 17 participating laboratories on two sets of multifilamentary NbTi strands (Set No. 1: copper matrix, fil. diam. between 0.5 and 12 mu m; Set No, 2: cupronickel matrix, fit, diam. between 0.4 and 1.2 mu m). The results reported here were measured by means of a.c. magnetization methods and separated into hysteresis and coupling losses. One laboratory used a calorimetric method. The data scatter in measured hysteresis losses among the participating laboratories was reasonably small for different measuring methods adopted and experimental arrangements used. On the other hand, the data scatter in coupling losses was large, mainly because in most laboratories a.c. losses were measured only at low frequencies (below 1 Hz), where the separation of coupling losses from total losses tends to be inaccurate. The comparison of measured hysteresis losses with the critical state model showed a large disagreement, which is assumed to be due to proximity effect coupling between filaments. (C) 1997 Elsevier Science Limited.