Methods are presented for the determination of copper in whole blood, plasma and serum using Zeeman effect flame and furnace atomic absorption spectroscopy. Three flame measurement modes were compared: continuous aspiration, microsampling in the peak height mode and microsampling in the peak area mode. The microsampler/peak area method was the most satisfactory. The precision for the microsampler/peak area method was as follows: the within-run relative standard deviation (RSD) was 1.84% and 1.89% for whole blood specimens with copper concentrations of 983 micrograms/l and 974 micrograms/l, respectively. The within-run RSD was 2.14, 1.66 and 0.87% for plasma specimens with concentrations of 990, 1,467 and 1,963 micrograms/l, respectively. Within-run RSD was 6.64, 2.86 and 1.15% for serum specimens with concentrations of 462, 984 and 2,056 micrograms/l, respectively. The average detection limit for the microsampler/peak area method was 10.3 micrograms/l. Concentrations could be read directly from standards prepared in human whole blood, plasma or serum pools or in a commercial control.