Background Plasmodium falciparum genotyping with molecular polymorphic markers is widely employed to distinguish recrudescence from re-infection in antimalarial drug efficacy monitoring programmes. However, limitations occur on agarose gel DNA measurements used to resolve the polymorphisms. Without empirical data, the current distinction of pre- and post-treatment bands, as persistent or new infection, is subjective and often varying by author. This study measures empirical tolerance limits for classifying different-sized bands as same or different alleles during MSP2 genotyping. Methods P. falciparum field samples from 161 volunteers were genotyped by nested PCR using polymorphic MSP2 family-specific primers. Data were analysed to determine variability of band size measurements between identical MSP2 alleles randomized into different agarose lanes. Results The mean (95% CI) paired difference in band size between identical alleles was 9.8 bp (1.48 – 18.16 bp, p = 0.022) for 3D7/IC and 2.54 (-3.04 – 8.05 bp, p = 0.362) for FC27. Based on these findings, pre- and post-treatment samples with 3D7/IC alleles showing less than 18 bp difference corresponded to recrudescence, with 95% confidence, while greater difference indicated new infection. FC27 allele differences were much narrower. For both 3D7/IC and FC27 amplicon, allele detection sensitivity was significantly higher with 13 μl compared to 20 μl or 30 μl lane loading volumes. Conclusion During MSP genotyping, it is useful to standardize classifications against measurement of background variability on identical alleles, in order to obtain reliable findings. It is critical to use a fixed optimal lane loading volume for constant allele patency, to avoid the disappearance or false appearance of new infection.