Sampling of platelets for trace-element analysis poses special problems: obtaining adequate sample material, achieving a sufficient cell purity, preserving viability (integrity), correcting for trapped plasma, and controlling contamination. We used a blood-cell separator for the primary isolation of platelets from blood, and differential centrifugation in natural plasma to further isolate them. The pyrimidopyrimidine RA233 was used as a stabilizer to maintain viability. 131I-labeled human serum albumin was used to estimate trapped plasma. Contamination was controlled by using five-times-distilled water to simulate donor's blood in the system and by comparing three fractions: the serum, the first portion of the platelet-rich plasma, and the supernatant plasma after the final centrifugation. Neutron activation analysis was used for the elemental analysis. A single differential centrifugation of the platelet-rich plasma from the blood-cell separator at 400 x g for 8 min was optimum (mean mass fractions:erythrocytes/platelets less than 5 mg/g and leukocytes/platelets less than 20 mg/g). The trapped plasma in the wet platelet samples amounted to about 0.40 g/g. No appreciable contamination from the sampling system was found for the elements Ag, Cd, Co, Cr, Cs, Cu, Fe, Mo, Rb, Sb, Se, and Zn.