Weed killer for driveways and patios: which one to buy

Every buying guide I read while writing this concluded that what a driveway needs is a systemic weedkiller that gets down to the roots. That is the right answer to the wrong question, and it is why people keep asking why the weeds are back by August.

A paving joint is not a plant problem. It is a seedbed. Grit, dust and windblown soil settle into the gap the jointing sand has washed out of, and seed lands on it all season. Killing today's dandelion perfectly, root and all, changes nothing about the seed that blows in next week and germinates in the same few millimetres of muck. That is the actual mechanism behind weeds coming back on a drive you have already sprayed.

A residual is the only chemistry that addresses it. Diflufenican, the residual in every UK path and drive product I could find, does not work through a living leaf the way glyphosate does. It inhibits an enzyme called phytoene desaturase, which the plant needs in order to make carotenoids, and carotenoids are what stop the light falling on a leaf from destroying its own chlorophyll. Block that and a seedling germinating through a treated layer comes up bleached white and dies before it establishes. Herbicide classification puts it in HRAC group 12, still widely labelled F1, and its database entry describes it in six words: selective contact and residual action, bleaching.

So the two halves of a path and drive bottle do genuinely different jobs on genuinely different plants. The glyphosate deals with what is standing today. The diflufenican deals with what has not germinated yet. If you take nothing else from this page, understand that the second one is what you are buying, because the first is available on its own for a fraction of the price.

Start with the thing that reframes the whole category, and it comes off the regulator's own register rather than anybody's opinion.

HSE keeps a public register of every pesticide authorised for amateur use in Great Britain. There are about 331 herbicides on it. The number containing diflufenican is four, all of them Pathclear, all of them Evergreen Garden Care, and every one is approved for the area "areas not intended to bear vegetation, unwanted vegetation". None is approved for hard surfaces, which HSE defines separately as any man-made impermeable surface such as concrete or asphalt that is not intended to bear vegetation. That is the gap: bare ground, gravel and permeable surfaces sitting on soil are one regulatory category, and your block-paved drive is another. The labels then say it outright, ruling out concrete, tarmac, decking and paved areas because the solution might run off onto adjacent land or into drains and gutters. That is the whole of the answer to whether path and drive weedkiller works on block paving, and I am not going to repeat it here.

This is not a quirk of the shelf product either. BASF's Pistol is the industrial version of the same combination, 250 g/L glyphosate with 40 g/L diflufenican under MAPP 18854, and its label is blunter than any amateur one. Pistol must not be used on non-porous man-made surfaces, and it gives paved areas, concrete and tarmac car parks and footpaths as its own examples. It may be used on porous surfaces such as gravel only where the underlying surface is soil, and not if an impermeable membrane lies between the gravel and the soil. And, in capitals, it must not be applied over drains or in drainage channels, gullies or similar structures for moving water.

Read the middle one twice, because it is the trap nobody flags. A gravel drive laid properly, over a woven membrane, is out of scope even for the professional product. The membrane is doing exactly what the restriction describes, sitting between the porous surface and the soil, so the residual has nowhere to bind and nowhere to go but sideways. Worth knowing, given that laying a membrane under gravel is the standard advice, mine included, in the honest breakdown of what a membrane does and does not stop.

Then the planting restriction, which is the biggest practical difference between the two product types. Glyphosate is inactivated the moment it touches soil, so Roundup Ultra's own page says you can replant the following day and the Optima+ concentrate says seven. The Pathclear label says allow at least six months, and dig the soil well first. Then it admits something manufacturers rarely put in writing: where this or other diflufenican products are applied in successive years, diflufenican will build up in the soil, and there is a risk of damage to subsequent plantings even if the soil is well dug. The professional label carries the same six-month interval and the same warning.

So the list is short. Never where the fall of the drive runs towards a border, a gully or a pond. Never over the root run of a tree, shrub or hedge you want to keep. Never on ground you might want to plant within the year, and think hard before a second successive year anywhere. Never on a day rain is forecast. And empty your washings well away from a drain, on ground you are not going to grow anything in.

Both labels, amateur and industrial, carry the same line at the bottom: very toxic to aquatic life with long lasting effects. That is not decoration. Diflufenican's toxicity to algae and to rooted aquatic plants is rated high, its chronic toxicity to fish is rated high, and in a water and sediment system its half-life in the sediment is around 373 days. Once it reaches water it stays there the better part of a year. For scale, the legal limit for an individual pesticide in UK drinking water is 0.1 micrograms per litre, and the Drinking Water Inspectorate names non-agricultural weed control alongside farming as a reason traces turn up in rivers and groundwater at all.

None of that framing is mine. Defra's own code of practice says it in a sentence: applying pesticides on hard surfaces may lead to run-off, take extra care that they do not enter drains or watercourses, and run-off can also pollute groundwater. And before a hard-surface weedkiller can be approved in Great Britain at all, amateur products included, the applicant must submit an assessment using HSE's HardSPEC model, which exists specifically to predict surface water, sediment and groundwater exposure from herbicide applied to hard surfaces. The run-off line on your bottle is there because a regulator modelled the run-off.

People ask constantly whether the stuff contractors use is stronger than the stuff in the garden centre. For this category I can answer with arithmetic off two approved labels rather than an opinion, and the answer surprised me.

Weedol Pathclear's maximum individual dose is 18ml of product per 20 square metres. Scale that to a hectare and it is 9 litres of product. At 20 g/L diflufenican and 125 g/L glyphosate, that puts 180 grams of diflufenican and 1,125 grams of glyphosate on the hectare.

Pistol, the industrial product, has a maximum individual dose of 4.5 litres per hectare. At 40 g/L diflufenican and 250 g/L glyphosate, that puts 180 grams of diflufenican and 1,125 grams of glyphosate on the hectare.

The same two numbers. Not close, identical. The trade pack is the shelf pack at exactly double the concentration applied at exactly half the volume, both cap you at one treatment a year, and both carry the same surface prohibition. Per square metre it comes out at 18 milligrams of diflufenican and 112 milligrams of glyphosate either way. So the difference is packaging, dilution and who is legally allowed to hold the lance, not potency. Professional describes the certificate the operator holds, not a stronger dose landing on the ground.

If you take one practical warning from this page, take this one, because it is the easiest mistake to make and it is about to get easier.

Weedol Pathclear is the residual: 125 g/L glyphosate plus 20 g/L diflufenican in the tubes, systemic and residual, one treatment a year, six months before replanting. Weedol Path & Gravel Control is 43.1 g/L pelargonic acid. No glyphosate, no diflufenican, no root kill, no barrier. Its own label states it will not provide long term control of perennial weeds, it allows up to four applications a year, and you can replant after three days. Two products, near-identical branding, opposite jobs, and the second is actively replacing the first on the shelf.

Resolva Path & Drive is worse, because it is not one product at all. There are three registrations under that name across two completely different actives: the original at 60 g/L acetic acid, a ready-to-use at 31.02 g/L pelargonic acid, and a concentrate at 186.7 g/L pelargonic acid. Same name on the front, and the only way to tell which one is in your hand is the MAPP number on the back.

The collisions keep going. Roundup Weedkiller Path Optima is the same pelargonic acid at the same 43.1 g/L as the Weedol product, because Evergreen Garden Care owns both brands. Neudorff WeedFree Plus and Vitax Ultima are the same formulation under two names, both authorised to Neudorff. So the fix is the one I give for every product question on this site: ignore the front of the pack, look for the active ingredient, and treat the MAPP number as the only thing that actually identifies what you are buying.

One more warning aimed squarely at buying any of this online. The retailer spec boxes for this category are not reliable and I would not decide anything from them. Checking current listings against the manufacturers' own labels, one Pathclear bottle is listed as covering approximately 50 square metres where its own label's maximum dose of 30ml per square metre puts a litre at 33. One Roundup listing gives its active ingredient as "green". A Pathclear listing gives its active ingredient as ferrous sulphate, which is not in it. The label on the bottle is the only document that has to be true.