A recent study published in Nature Communications analyzed 453,366 clinical trials to calculate success rates across phases, therapeutic areas, and time periods. The dataset represents one of the most comprehensive analyses of clinical trial outcomes to date.
The data offers a detailed view of where drugs succeed and fail in the development pipeline. Below, we summarize the key findings and what they suggest about the current state of pharmaceutical R&D.
Average development time
Average R&D cost per drug
Overall success rate (Phase I → Approval)
The Numbers at Each Phase
Let's start with what the data tells us about success rates at each stage of development. These numbers represent the probability of a drug advancing from one phase to the next:
Success Rates by Development Phase
Probability of advancing from one phase to the next
Source: Zhou et al., Nature Communications (2025). Analysis of 453,366 clinical trials.
The most striking finding? Phase II is where drugs go to die. With a success rate of just 32.4%, it represents the most challenging hurdle in drug development—often called the "valley of death."
The Funnel: From 10,000 to 1
To understand the full picture, we need to look at cumulative attrition. For every 10,000 compounds that enter the drug discovery pipeline, how many actually make it to patients?
The Drug Development Funnel
For every 10,000 compounds, approximately 1 reaches patients
Typical attrition rates in pharmaceutical R&D
This visualization makes the challenge visceral. The pharmaceutical industry isn't just facing a few percentage points of inefficiency—it's navigating a process where 99.99% of initial compounds never reach patients.
Success Varies Dramatically by Disease Area
Not all therapeutic areas are created equal. The data reveals striking differences in success rates across disease categories:
| Therapeutic Area | Phase II Success | Phase III Success | LOA (Phase I→Approval) |
|---|---|---|---|
| Hematology | 42.5% | 66.4% | 22% |
| Infectious Disease | 38.2% | 64.8% | 19.1% |
| Ophthalmology | 33.1% | 57.3% | 14.8% |
| Cardiovascular | 24.7% | 49.2% | 8.4% |
| CNS / Neurology | 18.9% | 47.1% | 6.2% |
| Oncology | 15.4% | 43.9% | 5.3% |
LOA = Likelihood of Approval. Source: ClinSR.org database analysis.
Hematology and infectious disease lead the pack, with Likelihood of Approval (LOA) rates of 22% and 19% respectively. Meanwhile, oncology—despite receiving the most R&D investment—has one of the lowest success rates at just 5.3%.
Why such variation? Several factors contribute:
- Biological complexity: CNS disorders involve the blood-brain barrier and complex neural circuits
- Endpoint clarity: Infectious diseases have clear, measurable outcomes (pathogen elimination)
- Patient heterogeneity: Cancer is really hundreds of different diseases
- Regulatory pathways: Some areas have more established approval frameworks
Key Findings from the Research
Key Findings
Orphan drugs show higher success
Drugs targeting rare diseases have significantly higher approval rates, benefiting from smaller trial requirements and supportive regulatory pathways.
Biomarkers boost approval odds
Trials using biomarker-based patient selection demonstrate 2-3x higher success rates, validating precision medicine approaches.
Success rates are improving
Overall clinical trial success rates have increased over the past two decades due to better trial design and patient selection.
Phase II remains the bottleneck
With only 32.4% success, Phase II is where most drugs fail—the critical transition from safety to efficacy demonstration.
The Timeline Problem
Beyond success rates, there's another critical dimension: time. The average drug takes 10-15 years from initial discovery to market approval.
Typical Drug Development Timeline
Discovery
2-4 yrs
Target ID
Preclinical
1-2 yrs
Lab & animal
Phase I
1-2 yrs
Safety
Phase II
2-3 yrs
Efficacy
Phase III
2-4 yrs
Confirmation
Review
1-2 yrs
FDA/EMA
Total: 10-15 years from discovery to market
This timeline has remained stubbornly consistent despite decades of technological advancement. While we've made progress in some areas (genomics, computational modeling), the regulatory and clinical trial processes have largely maintained their duration.
Implications for Regulatory Operations
Given these success rates and timelines, where can efficiency gains realistically be made? Clinical trial duration and regulatory review times are largely determined by scientific and procedural requirements. However, document preparation—the compilation of Clinical Study Reports, CTD modules, and supporting documentation—represents a significant time investment that is more amenable to optimization.
The regulatory submission process involves substantial manual work: cross-referencing across documents, ensuring consistency in terminology, compiling tables and figures, and formatting to regulatory standards. These tasks, while necessary, don't require the same scientific judgment as trial design or data interpretation.
This distinction matters when considering where automation and AI tools can appropriately be applied. Tasks that are rule-based, repetitive, and verifiable are candidates for automation. Tasks requiring scientific interpretation, regulatory strategy, or clinical judgment are not.
The question for regulatory teams isn't whether to use new tools, but which parts of the workflow can be made more efficient without compromising quality or regulatory acceptability.
📚 Reference
Zhou, Y., Zhang, Y., Xu, H. et al.. “Dynamic clinical trial success rates for drugs in the 21st century.” Nature Communications (2025).
DOI: 10.1038/s41467-025-64552-2
Read the full paper →