The main goal of non-invasive genetic capture-mark-recapture (CMR) analysis is to gain an unbiased and reliable population size estimate of species that cannot be sampled directly. The method has become an important and widely used tool to research and manage wildlife populations. However, researchers have to struggle with low amplification success rates and genotyping errors, which substantially bias subsequent analysis. To receive reliable results and to minimize the time and costs required for non-invasive microsatellite genotyping, one must carefully choose a species-specific sampling design, methods that maximize the amount of template DNA, and methods that could overcome genotyping errors, especially when using low-quality samples. This article reviews the literature and the pros and cons of the main methods used along the process described above. The review is strengthened by a case study on Eurasian otters (Lutra lutra) using feces; we tested several methods for their appropriateness to accommodate for genotyping errors. Based on this method testing, we demonstrated that high genotyping error rates are the key problem in this process leading to a severely flawed dataset if no consensus genotype is formed. However, even if generating consensus genotypes minimizes errors dramatically, we show that it may not achieve a definite eradication of all errors, which results in overestimated population sizes if conventional estimators are used. In conjunction with these findings, we offer a step-by-step protocol for non-invasive genetic CMR studies to achieve a reliable estimate of population sizes in the presence of high genotyping error rates.