A subdivision of the pollen types encountered in Lecythidaceae is proposed. The presence of a demarcation line between an original colpate and a derived syncolpate pollen type is confirmed. The significance of pollen characters for taxonomic subdivision is evaluated and it is concluded that the subdivision proposed by Niedenzu in 1892 agrees best with the pollen evidence. Pollen morphology does not yet provide any clear indications of wider affinities of the family, except in a negative sense.