The relation between inbreeding depression and rate of self-fertilization was studied in nine natural populations of the annual genus Amsinckia. The study included two clades (phylogenetic lineages) in which small-flowered, homostylous populations or species are believed to have evolved from large-flowered, heterostylous, self-compatible ones. In one lineage the small-flowered species is tetraploid with disomic inheritance. Rates of self-fertilization were 25% to 55% in the four large-flowered, heterostylous populations; 72% in a large-flowered but homostylous population; and greater than 99.5% in the four small-flowered, homostylous populations, which produce seed autonomously. When present, inbreeding depression occurred in the fertility but not the survival components of fitness. Using a cumulative fitness measure incorporating both survival and fertility (flower number), we found inbreeding depression to be lower in the four very highly self-fertilizing populations than in the five intermediate ones. The Spearman rank correlation between inbreeding depression and selfing rate for the nine populations was -0.50, but was not statistically significant (P = 0.12). Inbreeding depression was greater in the two tetraploid populations than in the very highly self-fertilizing, diploid ones. Phenotypic stability of progeny from self-fertilization tended to be higher in populations with lower inbreeding depression. We conclude that levels of self-fertilization and inbreeding depression in Amsinckia are determined more by other factors than by each other. Estimates of mutation rates and dominance coefficients of deleterious alleles, obtained from a companion study of the four highly self-fertilizing populations, suggest that a strong relationship may not be expected. We discuss the relationship of the present results to current theory of the coevolution of self-fertilization and inbreeding depression.