Genetic variation in coding regions is of strong interest for biologists as it represents an important factor that drives evolution. To analyse polymorphic loci, researchers usually rely on commonly used typing techniques such as cloning, SSCP, DGGE or RSCA. However, there are potential pitfalls in screening multi-allelic templates, which are mainly the formation of sequence chimeras during PCR amplification, and mosaic sequences during cloning. One of the most challenging genomic regions to explore is the Major Histocompatibility Complex (MHC), which codes for peptide-binding proteins of the vertebrate's adaptive immune system and is well known for its exceptional polymorphism. We compared the effect of two different PCR amplification approaches in a study of the MHC class IIB genes of the three-spined stickleback (Gasterosteus aculeatus). One approach used standard PCR conditions and the other a combination of several measures to eliminate PCR artefacts. In both approaches, the amplicons obtained were cloned and sequenced. In the first, established approach, 24% of the clones represented artefacts, while in the second approach the number of artefacts were reduced ten-fold. Furthermore, it enabled easy differentiation between real alleles and artificial sequences. We also analysed the potential effects of such artefacts in genetic analysis and evolutionary interpretation, and found a slight reduction in the signature of positive selection and an increase in recombination events. Consequently, we strongly recommend to apply the new PCR approach described in this study when genotyping MHC or other polymorphic genes.