Identification at the species level between microscopy and PCR was initially congruent for 80.5% of the cases. A second blind retrospective reading allowed us to improve the congruence of the Plasmodium identification to 89%. Discrepancies between microscopy and PCR were of three kinds: (i) the misinterpretation of some old P. falciparum trophozoites as P. malariae trophozoites; (ii) the misinterpretation of P. ovale infections as P. vivax infections; and (iii) the failure to detect mixed infections when the co-infecting species was in very low numbers