Abstract: A general MOPSO algorithm was applied to ZDT1-4. Bias in the archive solutions was observed in the initialisation of the archive solutions. The bias continued until simulation end because a general MOPSO algorithm does not contain any explicit way to correct bias in its archive. Pareto dominance testing was discovered to be a main contributor to the bias. Bias was also introduced by the target's problem landscape. Crowding distance was used to try to correct the bias, but was not generally successful in treating the effects of landscape bias. It was however successful in treating the effects of bias in the density of archive solutions. Such techniques treat the symptoms of bias. With a better understanding of the reasons for algorithm behaviour, it may be possible to develop techniques that treat the cause of bias instead.