When genetic diversity among organisms was measured with molecular markers, the question of genome coverage was currently stressed out. In order to check if well-distributed, mapped AFLP markers were more efficient in assessing varietal identification of carrot accessions than randomly chosen markers, nine closely related genotypes were analysed. A software was developed to realise 1,000 random choices of 20 to 70 mapped or unmapped markers, offering numerous genome coverages. We statistically showed that taking into account marker position does not provide a better estimation of genetic distances. Moreover, in the case of carrot, we concluded that 60 AFLP markers offer the best compromise between the level of precision and minimal expense.