Seed dormancy is hypothesized to be a risk-spreading strategy that maximizes plant fitness. By spreading germination time of offspring among years, plants may not only vary the environment to which their offspring are exposed, but may also reduce potential fitness losses due to sib competition. The sibling competition hypothesis predicts that large families whose offspring are more likely to experience sibling competition should stagger germination over several seasons (i.e. have greater dormancy) to a greater extent than small families. This should be observed as a lower initial probability of germination for seeds from large families than those from small families. We examined the relationship between family size and germination behavior in the desert mustard Lesquerella fendleri. In a single population in central New Mexico, USA, we estimated family size and germination fraction of 189 randomly selected maternal plants. Although a median test was not supportive, the randomization test showed marginal support for the sibling competition hypothesis. An additional study of twelve large and twelve small families in each of three other populations also showed a trend (significant in one population) for larger families to have lower germination fractions than small families. These results suggest a degree of support for the hypothesis, given that so many selective forces interact to generate patterns of seed dormancy. Thus, we have not ruled out the hypothesis that large families reduce the likelihood of sib competition in good years by producing seed crops with initially low germination. Sibling competition as well as environmental heterogeneity may have influenced the evolution of seed dormancy in this system.