Pollination, fecundity, dispersal, establishment, and survival of Lithospermum caroliniense (Boraginaceae) were examined over five years on sand dunes of different ages at Miller Dunes, Indiana, USA, to determine the effect of various environmental factors on the relationship between reproduction and other life history stages. Lithospermum caroliniense is a self—incompatible, iteroparous, distylous herb that depends on cross—pollination by insects for fertilization. Level of pollination was highly variable from year to year. For each of the two floral morphs, a high proportion of pollen was derived from the other morph in years of abundant pollination. In years when pollen loads were extremely low, fruit—set was limited by pollinator activity. Even with abundant transfer of compatible pollen, fecundity of L. caroliniense averaged less than one nutlet per flower. Individuals were consistent in level of fruit production from year to year. Emergence of seedlings of L. caroliniense was greatest when propagules were covered by 1—4 cm of sand in experimental field and garden studies. Propagules exposed on the sand surface were unlikely to obtain sufficient moisture for germination, while hypocotyls of seedlings that germinated at depths of @>5 cm were incapable of elongating sufficiently to reach the sand surface. Most propagules germinated after the first winter, but results from experimental plantings showed that many propagules remained dormant until after the second winter. First—season seedling survival averaged 7—34%, except in 1981, when frequent rainfall and moderate temperatures doubled the percentage that survived. The highest survival was among seedlings emerging near Lake Michigan or on sites with a shallow water table. Mortality of 1— and 2—yr—old plants was low. Individual plants grew faster and flowered at younger ages at sites closer to Lake Michigan. At the site closest to the lake, °83% of the 1978 cohort flowered at age 3 or 4 yr. At greater distances from the lake, most members of the 1978 cohort had not flowered by 1983. Except for some members of the 1981 cohort established near the lake, few plants in younger cohorts had flowered by 1983. In light of the slow growth rates of 1978—1982 cohorts and of many marked adults, large individuals of L. caroliniense are probably very old. Heavy nutlets typically fell within 60 cm of the parent plant. Dispersal was primarily downslope of adults, since adult foliage was lodged in a downslope direction by the time of propagule maturation; dispersal by wind or rolling was limited. Nutlets of L. caroliniense were consumed in large quantities, apparently by rodents, since the shelled remains of nutlets were common around adults. Occasional dispersal of intact nutlets by herbivores could be important in colonization of recently formed sand dunes. Although L. caroliniense is clearly dependent on newly formed sand dunes or disturbances, colonization is rare because dispersal of propagules is extremely limited. Large propagules are essential for establishment, but are not conducive to dispersal and colonization. Long adult life—span facilitates occasional colonization. Despite traits favoring establishment, seedlings suffered high mortality, except in years of abundant and evenly distributed rainfall. Since adult mortality was generally low, the life history of L. caroliniense resembles the pattern in which high juvenile mortality favors evolution of iteroparity and a long life—span. However, in contrast to species that produce a small number of offspring in any one year, the indeterminate, modular growth form of L. caroliniense means that increasing numbers of propagules are produced over the lifetime of an individual, until habitat change results in decline of adults. The combination of life—history characteristics occurring in this species argues against strict application of current life—history categorizations.