Sequencing projects are using paired-end reads to compensate for the relatively short length of the reads obtained by some sequencing technologies such as Illumina. To take advantage of the resulting paired-end reads, we need software that can handle paired-end reads by merging their overlapping parts. There is an ever-growing plethora of these software tools, all trying to be a little better in their respective fields to improve the result. However, none of these tools employ parallelisation on Graphical Processing Units(GPUs) to speed up the merging of the paired-end reads. What we aim to achieve in this project is a respectable speedup compared to existing solutions when it comes to merging of paired-end reads. The speedup might also give an opportunity to make better calculations of the score to achieve a better result. For this, I have developed a GPU implementation of a paired-end read merger based on FLASH, which employs parallelization on the GPU. This implementation was tested against FLASH in both speed and accuracy. Although the GPU implementation cannot quite catch up to FLASH, simple optimizations would allow it to easily compete with it, and shows great potential for paired-end read merging on GPUs.