Traditionally, broadcast spawning and planktonic larvae have been considered the plesiomorphic ‘ground plan’ for the Polychaeta and other metazoan groups. To assess whether this reproductive mode is in fact ‘primitive’, the study of monophyletic groups with various reproductive modes should be informative. A large range of body sizes would allow testing the ideas that aspects of reproductive mode may be functionally constrained. The family Sabellidac is one such group, with sexual reproductive modes ranging from broadcast spawning to intratubular brooding to ovovivi-parity, and a body size range over more than five orders of magnitude. Sabellids have previously been the subject of detailed cladistic analyses (Fitzhugh 1989, 1991); here we introduce several new characters based on morphology of reproductive structures. Larval development in four brooding sabellid species is also described with the aim of introducing new characters for future systematic analyses. Our cladistic analysis of sabellid genera suggests that gonochorism and brooding of direct-developing larvae are plesiomorphic in the Sabellidae, with external fertilization and swimming larvae limited to apomorphie clades in the subfamily Sabellinae. The presence of sperm with elongate heads may be correlated with the presence of intratubular brooding, though an adequate causal explanation for this relationship can not yet be presented. The concept that ‘modified’ sperm must be derived from ‘primitive’ sperm is shown to be false, with ‘modified’ sperm being plesiomorphic for the Sabellidae, from which ‘primitive’ sperm is derived in apomorphic Sabellinae. All sabellids have lecithotrophic development and appear to be phylogenetically constrained in this regard. Data gathered on body size and reproductive variables in the Sabellidac suggests the following (when phylogenetic effects are not controlled): (1) egg number and total egg volume are significantly correlated with body size, with small animals having fewer, larger eggs than large animals; (2) individual egg volume is not correlated with body size; (3) reproductive mode is significantly correlated with body size; intratubular brooders tend to be small-bodied, whereas broadcast spawners are large. However when the effect of body size is controlled for, then (4) egg number, egg volume and total egg volume all vary significantly with reproductive mode. Broadcast spawners expel a large number of small eggs for a high total egg volurne. Intratubular brooders have a few relatively large eggs for a small total egg volume. When statistics arc performed using phylogenetically independent contrasts there is a significant correlation between total egg volume and body size but not for egg number and body size. The effect of non-independence (due to phylogeny) of our data needs to be more fully controlled in future analyses but methods of incorporating continuous data into cladistic analyses should also be investigated. We show that some predictions can be made about reproductive mode based on body size but ad hoc patterns of reproductive character-state transformation should not be made independent of empirical hypotheses of phylogenetic relationship. Further studies of this kind throughout the Annelida are needed to determine the plesiomorphic reproductive mode for the phylum.