The paper presents а comparison of analytical performances of inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma atomic emission spectrometry (ICP-AES) for trace analysis of high purity bismuth and bismuth oxide. Matrix effects in the ICP-MS and ICP-AES methods were studied as a function of Bi concentration, ICP power and nebulizer flow rate. For ICP-MS the strong dependence of the matrix effects versus the atomic mass of analytes was observed. For ICP-AES the minimal matrix effects were achieved for spectral lines of analytes with low excitation potentials. The optimum degree of sample dilution providing minimum values of the limits of detection (LODs) was chosen. Both methods let us to reach LODs from n·10−7 to n·10−4 wt% for more than 50 trace elements. For most elements the LODs of ICP-MS were lower in comparison to ICP-AES. Validation of accuracy of the developed techniques was performed by “added-found” experiments and by comparison of the results of ICP-MS and ICP-AES analysis of high-purity bismuth oxide.