What Happened to Clinical Drug Failure Rates Over the Decades?
Clinical drug development has historically been plagued by high failure rates, with approximately 90% of drug candidates failing to reach market approval after entering human trials, a trend that has largely persisted for decades. While challenges like lack of efficacy, safety concerns, and poor study design remain prevalent, recent advancements in artificial intelligence (AI) are beginning to show promise in improving early-phase success rates and accelerating discovery timelines as of mid-2026. The industry is currently navigating a period of both persistent hurdles and transformative technological shifts.
Quick Answer
Clinical drug failure rates have consistently remained high over the decades, with an average of 90% of drug candidates failing to achieve regulatory approval after entering clinical trials. The primary reasons for these failures include insufficient efficacy, unmanageable toxicity, and flawed study design. However, as of 2026, artificial intelligence is emerging as a significant game-changer, demonstrating improved success rates in early clinical phases and compressing discovery timelines, although no AI-discovered drug has yet received full FDA approval. The pharmaceutical industry is actively integrating AI to enhance R&D productivity and address the long-standing challenges of drug development.
📊Key Facts
📅Complete Timeline12 events
Persistent High Failure Rates Documented
Studies analyzing drug development data from this period consistently reported an overall clinical trial success rate (Phase I to approval) of around 10-14%, with oncology drugs having significantly lower rates (3.4%). Phase II remained the biggest hurdle, with success rates often below 35%.
BIO Report Highlights Continued Challenges
A BIO report covering this decade found the overall likelihood of approval from Phase I for all developmental candidates was 7.9%. Phase II development remained the largest hurdle, with only 28.9% of candidates progressing, and programs with patient preselection biomarkers showed a two-fold higher likelihood of approval.
Lurbinectedin Receives Accelerated Approval, Later Fails Confirmatory Trial
Jazz Pharmaceuticals' lung cancer drug lurbinectedin (Zepzelca) received accelerated FDA approval. However, a confirmatory Phase 3 trial in 2026 failed to meet its primary goal of improving overall survival, highlighting the risks of accelerated approvals without robust long-term data.
Amylyx's Relyvrio Approved, Later Withdrawn
Relyvrio for ALS was approved based on positive Phase 2 results. However, a post-approval trial in 2024 showed it fared no better than placebo, leading Amylyx to withdraw the drug from the market in April 2024, demonstrating the challenges of validating efficacy in larger trials.
Increase in Completed Clinical Trials
Trialtrove recorded 4,295 industry-sponsored clinical trials from Phase I through Phase III/IV that reached completed status or reported primary endpoints, a 10.7% increase over 2022, reflecting a return to growth post-pandemic disruptions.
Several Major Clinical Trial Flops Reported
2025 saw significant late-stage failures for major players like Johnson & Johnson/BMS (Milvexian in Phase III), Pfizer (Inclacumab in Phase III), and Biohaven (Troriluzole in Phase III), primarily due to lack of efficacy or unacceptable side effects, leading to asset impairments and pipeline overhauls.
FDA on Pace for Lowest Drug Approvals Since 2022
By October 2025, the FDA had approved 42 new drugs and biologics, with projections for the full year indicating the lowest number of approvals since 2022 (51 approvals), potentially due to various factors including staff changes.
AI-Originated Drugs Show High Phase I Success Rates
As of early 2026, over 173 AI-originated drug programs were in clinical development, with AI-discovered molecules demonstrating an 80-90% success rate in Phase I trials, significantly exceeding the historical average of ~52%.
First AI-Designed Drug Completes Phase IIa with Positive Results
Insilico Medicine's AI-designed drug for idiopathic pulmonary fibrosis completed Phase IIa trials, showing dose-dependent improvement in lung function. This marked the first clinical proof-of-concept for an end-to-end AI-discovered drug, developed in 18 months at a cost of ~$6 million.
GLP-1s Drive R&D Return Improvement Amidst Broader Pressures
Deloitte's 'Measuring the return from pharmaceutical innovation' report noted that while overall R&D returns improved to 7.0% in 2025, this was largely driven by mega-blockbuster GLP-1 programs, masking persistent R&D pressures and increasing risk from single program failures.
FDA Approves 24 New Drugs by Mid-Year 2026
By July 7, 2026, the FDA's Center for Drug Evaluation and Research (CDER) had approved 24 new molecular entities and biological therapeutics, on par with mid-year levels in 2023 and 2024. Eli Lilly's oral obesity drug Foundayo was a noteworthy approval in April 2026.
Recent Oncology Trial Failures Highlight Ongoing Challenges
Several late-phase oncology trials failed in mid-2026, including a Phase 3 trial for lurbinectedin in lung cancer and Regeneron's immunotherapy combination in melanoma, underscoring the persistent difficulties in cancer drug development despite advancements.
🔍Deep Dive Analysis
The landscape of clinical drug development has been characterized by persistently high failure rates for many decades, making it one of the most challenging and costly endeavors in scientific research. Historically, around 90% of drug candidates that enter human clinical trials ultimately fail to gain regulatory approval. This high attrition rate translates into an average cost exceeding $2.8 billion and a timeline of 10-15 years for each successful drug brought to market. Despite numerous strategies implemented over the past 30 years, the overall success rate has remained stubbornly low, hovering between 10-15% from Phase I to approval.
The reasons for these failures are multifaceted and often occur at different stages of development. Lack of clinical efficacy is the predominant cause, accounting for 40-56% of Phase II and Phase III failures, where drugs show promise in preclinical studies but fail to demonstrate significant therapeutic effect in humans. Safety concerns and unmanageable toxicity are another major factor, contributing to approximately 28-30% of trial terminations. Other significant contributors include poor study design, inadequate patient recruitment and retention, insufficient funding, data management issues, and challenges in regulatory and ethical compliance. Phase II is often considered the 'valley of death' in drug development, exhibiting the lowest success rate of any phase transition, typically around 28-35%.
Clinical trial success rates also vary dramatically across therapeutic areas. Oncology drugs consistently face the lowest probability of success, with rates as low as 3-7% from Phase I to approval, reflecting the complex biology of cancer. In contrast, areas like rare diseases, hematology, and vaccines have historically shown higher success rates, sometimes reaching 25-33.4%. A key turning point in improving success has been the increasing use of patient preselection biomarkers, which can nearly double the likelihood of approval for development programs, raising it to 15.9% compared to 8% for unselected programs. This precision medicine approach helps in identifying the right patient populations for specific treatments.
As of 2026, artificial intelligence (AI) has emerged as a transformative force, offering a tantalizing glimpse into a future with improved R&D productivity. AI is being increasingly integrated across the drug development pipeline, from target identification to clinical trial design. Projections indicate that AI-enabled workflows can compress early discovery timelines by 30-40% and reduce preclinical candidate development to 13-18 months, significantly faster than traditional methods. While no AI-discovered drug has yet received full FDA approval, the first such milestone is projected for 2026-2027 with approximately 60% probability. AI-discovered molecules have already demonstrated significantly higher success rates in Phase I trials (80-90%) compared to the historical average of around 52%. By June 2026, the AI drug discovery market had grown to $4.2 billion globally, with over 173 AI-originated drug programs in clinical development.
Despite the promise of AI, the industry continues to face significant headwinds in mid-2026, including rising costs, increasing trial complexity, and patient recruitment challenges. The FDA approved 46 new drugs in 2025, a slight decrease from previous years, and as of mid-July 2026, 24 new molecular entities and biological therapeutics have been approved by the FDA's CDER. Recent high-profile clinical trial failures in 2025 and 2026, particularly in oncology and other complex diseases, underscore the ongoing difficulties, even for major pharmaceutical companies. However, the shift towards quality over quantity in clinical trials and the strategic adoption of AI are expected to lead to stronger outcomes in the coming years, with an estimated 70-80 novel active substances expected to launch annually over the next five years.
What If...?
Explore alternate histories. What if Clinical Drug Failure Rates Over the Decades made different choices?