IP67 means the fixture works after you take it out of the water. Nobody takes an in-grade light out of the ground.

Understand what IP67 actually promises

IP67 is dust-tight plus immersion to one meter for thirty minutes, and the pass condition is that the unit works after removal. That is a rating about surviving water that goes away, which is exactly right for a fixture that occasionally sits in standing water after a storm and exactly wrong as an assumption about a fixture permanently sitting in a wet hole. Guidance confirms the intended application, describing in-grade installations as susceptible to standing water after heavy rain or snowmelt with the immersion test directly simulating that risk. So specify IP67 for in-grade work, and then make the site behave the way the rating assumes it does.

Drainage is what makes the rating true, so buy the position first

The cheapest drainage decision on any in-grade installation is where the fixture goes. Guidance on enclosures asks the question in a transferable form: can you raise the enclosure or add drip shields to reduce immersion time? On an in-grade the equivalent is whether the surrounding grade sheds water away or funnels it in, and a fixture six feet along a wall in a spot that drains is a completely different proposition from the same fixture at the low point of a lawn. Walk the property during or just after rain and watch where water actually collects, because a dry afternoon tells you nothing and every plan drawn on a dry afternoon puts fixtures in puddles. Then build the drainage bed the fixture instructions call for rather than dropping the housing into a hole.

A bigger IP number is not automatically better

This surprises people and it is worth knowing before paying a premium. The ratings are not a ladder: they are not cumulative beyond IP66, and an IP67 product has been tested for immersion but not for high-pressure water jets, while an IP66 product has been tested for jets and not for immersion. They are different tests covering different risks, and higher is not always better because it depends on the exposure. For an in-grade set into a patio that both pools after rain and gets pressure washed twice a year, neither number on its own describes the exposure, and the answer is to check whether the specific fixture has been tested for both rather than assuming a larger number covers a smaller one.

The IP number excludes almost everything that kills an in-grade

This is the most useful sentence in the whole subject. An IP rating does not prove corrosion resistance, UV resistance, thermal performance, impact rating, electrical safety, or whether the light is listed for a specific wet location under electrical code. Now read that list against what an in-grade fixture endures: it is buried in soil, which is corrosion. It is walked on, which is impact. It is a sealed housing with a light source inside and earth packed around it, which is thermal. And it is in a wet location, which is a code listing question. The specification people shop on covers none of those four, so the IP number belongs on a checklist rather than at the end of one. The NEMA comparison makes the gap concrete: IP67 is close to NEMA 6 for temporary submersion, but NEMA ratings also cover corrosion resistance and IP ratings do not.

Heat has nowhere to go, and the rating says nothing about it

One of the four questions the enclosure guidance says to ask before choosing a rating is how much heat the equipment generates, and it matters more for a sealed in-grade than for anything else in a landscape. Every other fixture on a property sheds heat into moving air. An in-grade sheds it into a housing surrounded by soil, and thermal performance is explicitly outside what the IP rating covers. The practical consequences are two: take seriously any lamp or wattage limit the manufacturer states, because it is a thermal limit rather than a suggestion, and remember that a fixture set into paving where people walk barefoot has a surface temperature worth asking the manufacturer about rather than assuming.

The rating is only as good as the cable entry

A specific and easily missed point: all cable glands, vents and drains must be equally rated to maintain IP67 or IP68. The fixture is a system, not a housing, and the weakest rated component sets the actual protection. That means using the gland supplied rather than substituting, matching it to the actual cable diameter rather than approximately, and not improvising a cable entry in the field. Guidance is blunt about the broader version of this: the rating solves the immersion issue, but poor cable management, drainage or material selection will undermine it, so do not rely on IP67 alone. A perfect fixture with a sloppy entry is an unrated fixture.

Check the load rating separately, and specify it if vehicles are involved

Impact rating is another thing outside the IP number, and in-grade fixtures are the only landscape fixtures routinely stepped on. A fixture rated for foot traffic in a lawn or a walkway is a different product from one rated to be driven over in a driveway or a parking area, and the difference is a structural specification rather than a finish option. Get the fixture’s own load rating from its documentation and match it to the traffic the location will actually see, including the occasional case rather than the normal one, since a driveway apron gets a delivery truck sooner or later. Where vehicles are a possibility at all, specify a drive-over rated fixture rather than hoping.

Ignore the word waterproof, and test before backfill

Two practical habits. First, treat waterproof as marketing rather than specification: guidance notes the word is used loosely and that only fixtures with an immersion rating, usually IP67 or IP68, should be considered for water exposure beyond rain or spray. If a listing says waterproof without a number, it has told you nothing. Second, power the fixture and confirm it works before the hole is backfilled and the grade is restored, for exactly the reason the hardscape page gives about capping a wall: the moment before you close it up is the last moment it is cheap to reach. Then come back after the first heavy rain and look, because that is the only real test of whether the water leaves.