One peculiarity of protists belonging to classes Kinetoplastea and Diplonemea within the phylum Euglenozoa is compartmentalisation of most glycolytic enzymes within peroxisomes that are hence called glycosomes. This pathway is not sequestered in peroxisomes of the third Euglenozoan class, Euglenida. Previous analysis of well-studied kinetoplastids, the ‘TriTryps’ parasites Trypanosoma brucei, Trypanosoma cruzi and Leishmania spp., identified within glycosomes other metabolic processes usually not present in peroxisomes. In addition, trypanosomatid peroxins, i.e. proteins involved in biogenesis of these organelles, are divergent from human and yeast orthologues. In recent years, genomes, transcriptomes and proteomes for a variety of euglenozoans have become available. Here, we track the possible evolution of glycosomes by querying these databases, as well as the genome of Naegleria gruberi, a non-euglenozoan, which belongs to the same protist supergroup Discoba. We searched for orthologues of TriTryps proteins involved in glycosomal metabolism and biogenesis. Predicted cellular location(s) of each metabolic enzyme identified was inferred from presence or absence of peroxisomal-targeting signals. Combined with a survey of relevant literature, we refine extensively our previously postulated hypothesis about glycosome evolution. The data agree glycolysis was compartmentalised in a common ancestor of the kinetoplastids and diplonemids, yet additionally indicates most other processes found in glycosomes of extant trypanosomatids, but not in peroxisomes of other eukaryotes were either sequestered in this ancestor or shortly after separation of the two lineages. In contrast, peroxin divergence is evident in all euglenozoans. Following their gain of pathway complexity, subsequent evolution of peroxisome/glycosome function is complex. We hypothesize compartmentalisation in glycosomes of glycolytic enzymes, their cofactors and subsequently other metabolic enzymes provided selective advantage to kinetoplastids and diplonemids during their evolution in changing marine environments. We contend two specific properties derived from the ancestral peroxisomes were key: existence of nonselective pores for small solutes and the possibility of high turnover by pexophagy. Critically, such pores and pexophagy are characterised in extant trypanosomatids. Increasing amenability of free-living kinetoplastids and recently isolated diplonemids to experimental study means our hypothesis and interpretation of bioinformatic data are suited to experimental interrogation.