Life history characteristics of two competing starfish (the large Pisaster ochraceus and the smaller Leptasterias hexactis) are examined with respect to their relevance to the predictions and "correlates" of density—independent and density—dependent, or r— and K—selection, theory. Interspecific comparisons indicate that use of fecundity as an indication of reproductive output is inadequate because there species have evolved radically different reproductive strategies (Pisaster females broadcast many, small eggs; Leptasterias females brood few, large fertilized eggs). Further, since similar proportions of their respective total body weights are allocated to reproduction, this measure of reproductive output is also inadequate. However, the energy content of gonads and the part of the storage organs used during reproduction of the smaller species is higher than that of the larger species, suggesting reproduction is indeed costlier to the former. Intraspecific comparisons between fecundity and energy committed to reproduction by Leptasterias at several areas indicates these characters are not directly altered by interspecific competition but are strongly affected by the density—independent effects of wave action. Interspecific competition affects fecunidity only indirectly by reducing body size of Leptasterias (body size and fecundity are positively correlated). Although density—independent factors are believed to increase fecundity and reproductive efforts, increased severity of wave action apparently decreases them. This evidently results from physical limitations imposed by the smaller species' strategy of brooding its young. The difference in cost of reproduction to each starfish apparently conforms to the notion that species can be ranked along a postulated "r—K continuum." However, fecundity of Leptasterias is reduced by increased density—independent mortality. Although no similar departures from r—K selection theory have been reported, the habit of brooding is fairly common in nature. Hence Leptasterias' response to wave action is probably not unique. These data at least suggest that the generality of r— and K—selection theory should be determined before theoretical elaboration of this theory is safe.