Previously,α-mannosidases were classified as enzymes that process newly formed N-glycans or degrade mature glyco-proteins. In this review, we suggest that two endoplasmic reticulum (ER) α-mannosidases, previously assigned processing roles, have important catabolic activities. Based on new evidence, we propose that the ER cytosolic manno-sidase is involved in the degradation of dolichol intermediates that are not needed for protein glycosylation, whereas the soluble form of Man9-mann0sida.se is responsible for the degradation of glycans on defective or malfolded proteins that are specifically retained and broken down in the ER. The degradation of oligosaccharides derived from dolichol intermediates by ER cytosolic manno-sidase now explains why cats and cattle with α-mannosidosis store and excrete some unexpected oligosaccharides containing only one GlcNAc residue. Similarly, the action of ER cytosolic mannosidase, followed by the action of the recently described human lysosomal α(l-6)-mannosidase, together explain why aαmannosidosis patients store and excrete large amounts of oligosaccharides that resemble biosynthetic intermediates, rather than partially degraded glycans. The relative contributions of the lysosomal and extra-lysosomal catabolic pathways can be derived by comparing the ratio of trisaccharide Manη(l→4)GlcNAc η(l→4)GlcNAc to disaccharide Manη(l→4)GlcNAc accumulated in tissues from goats with η-mannosidosis. A similar determination in human p-mannosidosis patients is not possible because the same intermediate, Manη(l→4)-GlcNAc is a product of both pathways. Based on inhibitor studies with pyranose and furanose analogues, α-mannosidases may be divided into two groups. Those in Class 1 are (1→2)-specific enzymes like Golgi mannosidase I, whereas those in Class 2, like lysosomal a-mannosidase, can hydrolyse (l→2), (l →3) and (l→6) linkages. A similar classification has recently been derived by others from protein sequence homologies. Based on this new classification of the α-mannosidases, it is possible to speculate about their probable evolution from two primordial genes. The first would have been a Class 1 ER enzyme involved in the degradation of glycans on incompletely assembled or malfolded glycoproteins. The second would have been a Class 2 lysosomal enzyme responsible for turnover. Later, other α-mannosidases, with new processing or catabolic functions, would have developed from these, by loss or gain of critical insertion or retention sequences, to yield the full complement of α-mannosidases known today.