In 2008, the International Measurement Evaluation Programme, IMEP®, organised two proficiency tests (PTs) for the determination of heavy metals in mineral feed, named IMEP-105 and IMEP-27, respectively. IMEP-105 was organised in support of the activities of the European Union Reference Laboratory for Heavy Metals in Feed and Food, and participation was restricted to the officially nominated National Reference Laboratories (NRLs). IMEP-27 was open to all interested control laboratories in the field. The test material used in the two PTs was the same and the timeframe of the two exercises overlapped. The measurands in both exercises were total Cd, Pb and As and extractable Cd and Pb, as defined in Directive 2002/32/EC of the European Parliament and the Council on undesirable substances in animal feed. Forty-nine laboratories from 25 countries reported results to the two exercises, 29 to IMEP-105 and 20 to IMEP-27. In both PTs, external reference values were used instead of consensus values derived from the participants' results. It was shown that the concentration of total and extractable Cd according to Directive 2002/32/EC were identical, while the concentration of extractable Pb was ~80% of total Pb. The main outcome of these comparisons was that an underestimation of the concentrations of the addressed measurands, in particular total Pb, took place due to an erroneous estimation of the bias of the analytical methods used by most of the participants. It appeared that the nature of the certified reference materials chosen for method validation and recovery estimation was the cause of the problem (insufficient matching of the matrix characteristics between these materials and the test sample). No significant difference between the results reported to IMEP-105 and to IMEP-27 was observed.