Primates are quite unique among placental mammals in that the two extreme types of placentation are present within a single order. Strepsirrhines (lemurs and lorisiforms) have non-invasive epitheliochorial placentation, whereas haplorhines (tarsiers and higher primates) have highly invasive haemochorial placentation. Resemblance in placenta type in fact provided the first evidence that tarsiers are linked to higher primates and distinct from lemurs and lorisiforms. Tree-shrews differ from both primate subgroups in having moderately invasive endotheliochorial placentation, while colugos have invasive haemochorial placentation like haplorhines. All three kinds of placentation have been identified as primitive for placentals by different authors, but until recently the prevailing interpretation has been that non-invasive epitheliochorial placentation is primitive and ''less efficient''. Opposing this interpretation, Martin (Primate origins and evolution: a phylogenetic reconstruction, 1990) proposed that moderately invasive endotheliochorial placentation is primitive. Epitheliochorial placentation is unlikely to be primitive because it is predominantly associated with large body size, relatively long gestation periods and precocial offspring. Furthermore, some strepsirrhines and other placental mammals with epitheliochorial placentation retain indications of former invasiveness of the placenta. The recent availability of comprehensive molecular phylogenies for placental mammals has provided an independent framework to determine the most parsimonious interpretation of the evolution of placenta types and other reproductive features. It has consistently emerged that epitheliochorial placentation is best explained as a derived condition, although opinions differ as to whether the ancestral condition for placental mammals (and hence for primates) was endotheliochorial or haemochorial. It is argued that on balance the most likely ancestral condition is endotheliochorial. Comparative evidence across placentals clearly indicates that epitheliochorial placentation is not less efficient than more invasive forms of placentation, at least with respect to growth in overall fetal body mass. The ratio of neonate mass to gestation period (a simple indicator of average daily maternal investment in fetal growth) shows no difference according to placenta type. Differential evolution of placentation is hence presumably linked to immunological factors, parent/offspring conflict and/or genomic imprinting.