Existing life-history concepts deal inadequately with clonal organisms. To make them more generally applicable, several assumptions must be widened: (i) life histories may be expressed at different levels of individual organization, e.g. genets and modules; (ii) parts of a life cycle may be executed more than once by a single genetic individual; (iii) genetic individuals do not necessarily senesce; (iv) size and reproductive value of genetic individuals may increase indefinitely with age; and (v) dormancy, dispersal of progeny, and production of genetically different progeny are not necessarily linked to the same process, as they are in reproduction of unitary organisms. A model is presented of a size-structured population of genets, in which each genet comprises an age-structured metapopulation of modules. The model predicts that the extent to which clonal growth evolves depends on: constraints in genet architecture (high or low within-genet `dispersal'); cost of reproduction relative to that of clonal growth; the presence or absence of density-dependent regulation of population size; the presence of lethal events that kill entire genets, as opposed to genet mortality occurring only as a consequence of the mortality of all its modules; the ability of genets to predict such lethal events; and the temporal heterogeneity of the environment. The predicted responses to r- and K-selection differ markedly from those predicted for unitary organisms, especially for genets with high within-genet dispersal. For example, r-selection may favour clonal growth with no reproduction, whereas K-selection favours greater reproduction.