The mechanism of matrix interference in ICP spectrometry has been probed by incorporation of the droplet-desolvation and solute-particle vaporization processes in a computer simulation of the ICP. Two types of simulation have been carried out: single-aerosol droplet studies, and many-droplet studies. The first type of calculation is used to examine the response of the vaporization kinetics to changes in plasma properties (central channel flow rate and forward power) and to aerosol properties (initial droplet diameter, concentration of interferent, and thermal and physical properties of the interferent). From these studies, the dominant mechanism for solute-particle vaporization in the ICP is small-particle mass transfer. In addition, the height in the plasma at which a solute-particle has completely vaporized is related roughly linearly to the initial diameter of the aerosol droplet, linearly to the central channel flow rate, inversely to the forward power, and to the square root of interferent concentration. The second type of calculation is a many-droplet simulation for direct comparison with experimental results. Ground state atom and ion number densities have been calculated by simulation of plasma properties, aerosol desolvation and vaporization, and flow and diffusion of the analyte vapor. These studies indicate that if early vaporization and lateral diffusion are dominant reasons for the interelement interference las experiments have suggested) then one (or both) of two things must also be true. First, the limiting mechanism of vaporization predicted by theory may be incorrect, since the small-particle mass-transfer rate constants for all interferents are lower than or equal to that for CaCl2, the analyte. Also, liberation of the EIE below the load coil may raise the gas temperature along the central channel, consistent with experimental results; this rise causes an increase in the rate of desolvation and possibly vaporization. (C) 1998 Elsevier Science B.V. All rights reserved.