Where the science genuinely runs out
The qualifying work is where a competent professional could not simply deduce the answer. In pharmaceuticals that is novel formulation to overcome poor solubility, bioavailability or stability; new delivery systems, from extended-release to transdermal, inhaled or drug-device combinations; and new synthetic routes or biologics manufacture where yield, purity or behaviour is genuinely uncertain. In devices it is engineering under conflicting constraints, miniaturisation against power, thermal and biocompatibility limits, novel sensors and implantables, sterilisation that does not degrade function. In diagnostics it is developing and validating a novel assay, improving sensitivity or specificity, or miniaturising a lab method to the point of care.
Bioinformatics and digital health qualify on the same footing as any software: a genuinely novel algorithm where the technical approach is uncertain can be R&D, but configuring or retraining an off-the-shelf model on new data is not.
The work that feels like R&D but is not
Much of the effort in a life-sciences company sits outside the definition, however skilled and expensive. Trials that only gather efficacy or safety data on a known compound by established protocols are data-gathering, not the resolution of technical uncertainty. Regulatory approval work, pharmacovigilance, bioequivalence studies for generics, brand-name research, health economics and market access are all excluded. So are routine quality control, analytical testing to a set method, and usability tweaks to a health app.
The clinical-trial boundary
This is the line HMRC looks hardest at, and the phase label does most of the misleading. Ask what the activity is actually resolving. A trial that tests a first-in-class mechanism, a novel formulation, or a drug-device interaction can be resolving genuine scientific uncertainty. A trial that confirms the safety and efficacy of a known approach, generating the data a regulator requires, is not, however demanding it is to run. R&D ends when the uncertainty is resolved; scale-up of a proven process and the march to approval and launch are not R&D.
Sector pitfalls we would check first
- The phase-equals-qualifying assumption. Neither “it was a clinical trial” nor “it was Phase III” settles anything. Each activity is tested against whether it directly contributes to resolving a specific scientific or technological uncertainty.
- Contracted-out clinical work. Where a CRO runs the work, the merged scheme generally gives the claim to the party that decided on and articulated the R&D, usually the sponsor, not the CRO. The contracts and contemporaneous evidence have to support that.
- The overseas restriction. For accounting periods from 1 April 2024, payments to overseas CROs and workers generally no longer qualify. The narrow exemption turns on conditions genuinely unavailable in the UK, such as a specific patient population, and never on cost or convenience. Multi-site trials have to be apportioned.
- Where the costs sit. Staff, consumables and reagents, clinical-trial volunteer payments, and data and cloud costs for bioinformatics all have their own rules; capital lab equipment sits outside the revenue claim and may belong in qualifying costs as a separate matter.
Frequently asked questions
Do clinical trials qualify?
Sometimes, and the phase is only a proxy. HMRC accepts that discovery, pre-clinical and Phase I to III work is usually concerned with resolving scientific or technological uncertainty, while Phase IV and post-marketing surveillance usually is not. The real test is whether the activity directly resolves a specific technical uncertainty, not which phase it sits in.
Does our regulatory and approval work count?
No. Preparing MHRA, CE or UKCA submissions, pharmacovigilance and bioequivalence studies for generics are not R&D in themselves. They are how you demonstrate and license work already done. The underlying development that created the uncertainty may qualify; the approval process does not.
We are loss-making and R&D-heavy. Is there a better route than the merged scheme?
Possibly. Loss-making SMEs whose R&D is a large share of total spending may claim through ERIS instead, which is often more valuable than the merged credit. Whether you qualify turns on a strict intensity test. See ERIS explained.
Does developing a new assay or diagnostic qualify?
It can, where you are developing or validating a genuinely novel method and the sensitivity, specificity or reproducibility is uncertain at the outset. Running an established assay to a standard protocol is routine testing, not R&D.