Centrioles are essential for the formation of cilia, flagella and centrosome organization. Abnormalities in centrosome structure and number in many cancers can be associated with aberrant cell division and genomic instability.(1,2) Canonical centriole duplication occurs in coordination with the cell division cycle, such that a single new "daughter" centriole arises next to each "mother" centriole. If destroyed, or eliminated during development, centrioles can form de novo.(3-5) Here we discuss our recent data demonstrating a molecular pathway that operates in both de novo and canonical centriole biogenesis involving SAK/PLK4, SAS-6 and SAS-4.(6) We showed that centriole biogenesis is a self-assembly process locally triggered by high SAK/PLK4 activity that may or not be associated with an existing centriole. SAS-6 acts downstream of SAK/PLK4 to organize nine precentriolar units, which we call here enatosomes, fitting together laterally and longitudinally, specifying a tube-like centriole precursor.(7,8) The identification of mutants impaired in centriole biogenesis has permitted the study of the physiological consequences of their absence in the whole organism. In Drosophila, centrioles are not necessary for somatic cell divisions.(9,10) However, we show here that mitotic abnormalities arise in syncytial SAK/PLK4-derived mutant embryos resulting in lethality. Moreover male meiosis fails in both SAK/PLK4 and DSAS-4 mutant spermatids that have no centrioles. These results show diversity in the need for centrioles in cell division. This suggests that tissue specific constraints selected for different contributions of centrosome-independent and dependent mechanisms in spindle function. This heterogeneity should be taken into account both in reaching an understanding of spindle function and when designing drugs that target cell division.