Interviews were carried out amongst a cohort of 488 parents who lived as subsistence farmers in a remote and very poorly resourced region of South Africa. The majority of respondents had invested heavily in the education of their children, contrary to what might have been predicted from large family sizes and the economic and ecological pressures that families faced. There was little evidence of child specialization, in which educational investment might have been targeted more to some children than others in a family. However, relatively wide birth spacing may have made the financial and opportunity costs of schooling more manageable. This speculation is given some support by two findings: first, that the number of grades of schooling children complete before leaving increases significantly as birth spacing increases; second, that children who are still in school progress more rapidly as birth interval increases. Greater opportunities for the schooling of sons (made possible by low rates of migrant labour in the community), coupled with high opportunity costs associated with the schooling of daughters, made it likely that sons would be educated to a higher level than daughters. However, there was consistent evidence for gender equity in schooling patterns. This is attributed to the lower risk of financial and opportunity losses that are associated with rearing daughters in this particular community. Birth order exerted only modest effects on the decisions parents made about schooling their offspring. The investment strategy as a whole could be interpreted as ``bet-hedging'' under conditions where events outside parental control prevented them from targeting investment to children who could be readily identified as having favourable educational and employment prospects. It is concluded that parents invest in their sons and daughters in a manner that can only be understood when the complexities of their particular social, economic, and ecological contexts are taken into account.