Produce a double stranded amplicon, both being undesirable for most oligonucleotide library applications. In the present work, we have compared three different approaches to remove primer binding sites and deliver libraries of single-stranded oligonucleotides, namely alkaline denaturation, exonucleolytic strand removal and in vitro transcription-reverse transcription. While alkaline denaturation is appealing for its simplicity, we have highlighted several drawbacks. First, the primer binding sites are removed by using nicking enzymes to specifically cleave the PBS of the desired strand. There is a very limited repertory of nicking enzyme, making it quasi impossible to design an oligonucleotide library omitting these recognitions sites. Second, the mild alkaline denaturing conditions used to melt the double stranded DNA, break the biotin – streptavidin bond to a significant extend, leading to the contamination of the desired product with complementary strands. While these strands could be removed by a second binding to beads, there is a risk that they have re-hybridized to a complementary strand. One could propose to first heat denature the PCR amplicons and then perform the removal of unwanted biotinylated strands with magnetic beads. This would work for complex libraries but will undoubtedly fail for low complexity ones, such as libraries of point-mutations of the same oligonucleotide coding for short polypeptides. Finally, in our hands alkaline denaturation method with a biotin-streptavidin affinity selection necessitated PAGE purification to get desired size ssDNA resulting in very low yields and we do not recommend it. For the generation of small amounts of oligonucleotide, exonucleolytic strand removal is the preferred approach. Lambda exonuclease degrades the phosphorylated strand with much greater affinity than the non-phosphorylated one. Alternative conclusions cannot be ruled out. Ideally, a library of individually synthesized oligonucleotides would be pooled at equimolar concentration and subjected to PCR then IVT-RT amplification and used for normalization. This approach is too expensive to be practical. Another approach would be to perform deep sequencing of the original and the amplified library. However current NGS platforms also come with inherent biases and preferential sequence drop-out during sequencing is possible. To summarize, applications requiring less than 1 mg of single stranded oligonucleotide libraries, we recommend using the exonucleolytic strand removal because of its simplicity. However, when larger amounts are requested, it becomes necessary to implement the in vitro transcription – reverse transcription method. It is important to note that the present reverse transcription method can also be followed by a second strand synthesis offering an appealing alternative to large scaling up of emulsion PCR amplification when double stranded libraries are considered.