This paper looks at the optimal location of new forests in a suburban region under area constraints. The methodology takes into account use benefits, non-use benefits (both bequest and existence values), opportunity costs of converting agricultural land, as well as planting and management costs of the new forest. The recreational benefits of new forest sites are estimated using function transfer techniques. We show that the net social benefit of the total afforestation project varies up to a factor 6. The recreational value of a site varies considerably with the available substitutes.