Does companion planting actually work?

Some of it does, with reasonable evidence. A great deal of what circulates is traditional lore that has not held up when tested, and the two are rarely distinguished in the advice you find.

What has decent support:

Trap cropping — growing a plant pests prefer to draw them off the crop. Nasturtiums attract cabbage white butterflies and blackfly away from beans and brassicas, and this is measurable.

Attracting beneficial insects. Planting umbellifers and daisies — dill, fennel, yarrow, calendula, phacelia — brings in hoverflies, lacewings and parasitic wasps whose larvae eat aphids. This is well supported and is the single most useful thing on the list.

Physical disruption. Interplanting so that a pest's target crop is not in an obvious monoculture block genuinely slows pests that locate plants visually. Undersowing brassicas with clover reduces cabbage root fly damage in trials.

Structural pairing — the Three Sisters of maize, beans and squash, where each contributes support, nitrogen or ground cover.

What is weaker: many specific pairings passed down as fact, such as the claim that basil improves the flavour of tomatoes, or long lists of plants that supposedly must never be grown near each other. Most have never been tested, and some tested claims have failed.

Allelopathy is real but narrower than claimed. Some plants genuinely inhibit others chemically — walnut trees release juglone, which harms many species nearby — but the sweeping "enemy plants" charts go well beyond the evidence.

The reliable principle underneath all of it is diversity: mixed planting supports more predators and is harder for pests to exploit than a block of one crop.