Designing a Garden Deer Will Not Eat
Five decisions, in order. Made at the start they cost nothing; made afterwards they turn into a spraying habit you keep for years.
Ceaser Sinclair · July 29, 2026 · 4 min read
Repellents and fencing manage deer damage. Design decides how much of it there is to manage.
That is not a slogan. Every published landscape design sequence puts plant selection near the end, after purpose, survey and layout. NC State's runs to six steps with plants at five. Deer make that ordering expensive to get wrong, because the plants you would have chosen last are the ones that determine whether you spend the next decade spraying.
These five decisions come in order. Each one narrows the next.
Purpose
What is the space for, and what would losing it mean?
This is first because it sets the budget. Minnesota Extension's benchmark is that a 50 per cent reduction in deer browse is very successful and 30 per cent is the likelier result. Thirty per cent is fine for a shrub border and useless for a cutting garden, and knowing which you are building tells you whether to buy plants or fence.
It also settles where things go. Rutgers' instruction for its bottom two tiers is that those plants should only be grown with additional protection, which makes the rating a rule about placement rather than just about purchase.
Structure
The bones: evergreens, hedges, walls, the shapes that carry the garden in February.
February is also when browse pressure peaks, so the planting you depend on in winter is the planting most exposed. The decision that matters is not which evergreen gets eaten but which one recovers. Conifers carry no dormant buds on old stems, so foliage stripped below the browse line never returns and the plant keeps that shape for life. Broadleaf evergreens like box and holly can rejuvenate.
Two plants can take identical damage in the same January, and one is disfigured permanently while the other is fine in two seasons.
Theme
Some themes are deer-resistant almost by accident, because the traits that suit a hot dry site are the traits that deter browsing.
There is a mechanism behind committing to one rather than dotting resistant plants about. Deer use plant volatiles as a pre-ingestive cue: after sniffing high-terpene plants they were reluctant to begin foraging at all. A bed that smells of lavender and thyme across its whole area is producing that signal continuously. One lavender among six hostas is not.
Colour
The resistant palette bunches into silver, blue and purple, and it does so because of the defences themselves. Aromatic oils come with the mint family and its blue-violet flowers. Silver foliage is what a dense layer of defensive hairs looks like.
Red is the genuinely hard part, because roses, daylilies and tulips own that range and all three sit in the half of the list that comes with a warning.
Contrast
The last decision, and the one that rescues a narrow palette or loses it.
Design theory says a composition needs one coarse texture as a focal point, with fine and medium filling around it. The deer-resistant palette is systematically short of coarse texture, because narrow aromatic leaves, felted silver foliage and small leathery evergreens all read as fine or medium. A border of nothing but those is pleasant and has no subject.
Why this order
Each decision removes options from the next, and doing them backwards is what produces the garden that has to be defended.
Choose plants first and purpose becomes a justification for what you already bought. Choose colour before theme and you go looking for red in the one part of the spectrum the palette cannot supply. Choose structure last and the bones end up being whatever was left, which is how arborvitae becomes the hedge on nine properties out of ten.
Applied at the start, none of this feels like a compromise, because the constraint shaped the design instead of arriving as a series of substitutions.
When you want a worked example rather than a principle, the free plans apply all five to specific hard sites.
Ratings referenced throughout come from the Rutgers list, which is practitioner consensus from northern New Jersey rather than a controlled feeding trial. It names no deer species, and mule and black-tailed deer browse differently.