Biotech Phase I II III Probability: Success Rates and rNPV

Key Takeaways

• Biotech Phase I II III probability: Phase I-to-approval averages roughly 7.9–10.4 percent cumulative per BIO and Tufts datasets, with the highest attrition at Phase II where the first real test of biological effect occurs.
• Therapeutic area matters enormously: oncology runs roughly 5–6 percent cumulative while hematology and rare diseases can exceed 15 percent; applying the industry average to any specific asset routinely misprices it.
• A common mistake is forgetting financing risk when compounding clinical PoS; a small-cap biotech with a Phase III readout in three years and 24 months of cash has an embedded dilutive financing event that must be modeled alongside clinical probability.
• A positive Phase II proof-of-concept readout can triple rNPV in a single day because cumulative PoS jumps from roughly 12 percent to roughly 44 percent for an oncology asset, explaining the binary gap-on-data behavior of biotech stocks.

What It Is

Phase probabilities of success (PoS) quantify the historical frequency that a drug candidate clears each clinical phase and ultimately receives marketing approval. The four standard transitions are Phase I to Phase II, Phase II to Phase III, Phase III to filing (NDA or BLA), and filing to approval. Multiplying the four gives the cumulative probability from first-in-human to commercial launch.

The Tufts Center for the Study of Drug Development and the BIO industry association both publish multi-decade datasets, with the BIO Clinical Development Success Rates 2011 to 2020 study being the most cited cross-industry reference. Cumulative success from Phase I to approval averages around 10 percent, with wide variation by therapeutic area.

The Intuition

A clinical pipeline is a series of low-probability bets. Most candidates die at Phase II for efficacy reasons, the so-called valley of death, because Phase I established safety in healthy volunteers but Phase II is the first real test of biological effect. Phase III then tests the same effect in a larger, more diverse population where placebo response and statistical noise can swamp a marginal drug.

Because most candidates fail, the candidates that survive must carry enough peak-sales potential to fund the failures. A platform with five Phase I assets at 10 percent each has roughly half a launch in expectation, which is why biotech valuation is so sensitive to single-asset readouts and why managements diversify aggressively.

How It Works

The basic decomposition uses conditional probabilities at each transition.

PoS(I to approval) = PoS(I to II) * PoS(II to III) * PoS(III to filing) * PoS(filing to approval)

Industry-aggregate values from BIO 2021 (covering 2011 to 2020) and Tufts CSDD studies cluster near these levels:

Phase I to Phase II:      ~52 percent
Phase II to Phase III:    ~29 percent
Phase III to filing:      ~58 percent
Filing to approval:       ~91 percent
Cumulative I to approval: ~7.9 to 10.4 percent depending on dataset and indication

Therapeutic-area variation is large. Hematology and rare diseases tend to show higher cumulative PoS (often well above 15 percent), while oncology has historically been the lowest performer at roughly 5 to 6 percent cumulative. Validated targets and biomarker-selected populations raise PoS materially.

A risk-adjusted net present value (rNPV) model multiplies peak unrisked sales by cumulative PoS and discounts cash flows over the launch curve.

rNPV = sum over t of [ (Sales_t * cumulative PoS at stage t) / (1 + r)^t ] - PV(remaining R&D)

Worked Example

Consider a hypothetical oncology asset entering Phase II with peak unrisked annual sales of 2.0 billion dollars and an expected launch in eight years. Use BIO oncology phase probabilities of approximately 28 percent (Phase II to III), 50 percent (Phase III to filing), and 88 percent (filing to approval). Cumulative remaining PoS from start of Phase II is:

PoS = 0.28 * 0.50 * 0.88 = 0.123 (12.3 percent)

Risk-adjusted peak sales: 2.0 billion times 0.123 equals 246 million dollars. Apply a launch curve that ramps over five years to peak, hold for several years, then declines after loss of exclusivity. Discount at a biotech-appropriate rate (10 to 12 percent is common for clinical-stage assets). The rNPV before subtracting remaining R&D might land in the 600 million to 1 billion range depending on assumptions.

If the asset reads out positive Phase II proof-of-concept, cumulative PoS jumps from 12.3 percent to roughly 50 percent times 88 percent equals 44 percent, and rNPV roughly triples. That non-linear revaluation is why biotech stocks gap on data.

Common Mistakes

1. Using a single industry-average PoS for every asset. A first-in-class oncology candidate without a validated target deserves a far lower PoS than a follow-on in a class with three approvals. Always condition on indication, mechanism, and trial design.

2. Ignoring the difference between investigational and approved-mechanism trials. Trials of approved mechanisms in new indications (label expansions) clear at materially higher rates than novel mechanisms in any indication. Tufts and BIO both report this gap, and analysts who skip the conditioning routinely overpay for novelty.

3. Compounding PoS without accounting for funding risk. A small biotech with two years of cash and a Phase III readout in three years has an embedded financing event. The risk-adjusted equity value must reflect dilution probability, not just clinical PoS.

4. Treating filing-to-approval as a near-certainty. The roughly 90 percent rate is real, but it includes complete response letters, REMS requirements, and label restrictions that can crater commercial value even when approval comes. Read the FDA briefing documents, not just the headline.

5. Anchoring on Phase III PoS for a confirmatory trial of an already-approved drug. Subgroup expansions and combination trials of approved agents have very different success patterns and should not be modeled with the standard 60 percent Phase III to filing rate.

Frequently Asked Questions

Q: What are biotech Phase I II III probabilities in simple terms? Phase probabilities of success measure how often drug candidates clear each clinical hurdle on the way to FDA approval. Per BIO and Tufts data: roughly 52 percent of Phase I assets advance to Phase II, 29 percent of Phase II assets advance to Phase III, 58 percent of Phase III assets reach filing, and 91 percent of filed NDA/BLAs receive approval. Multiply those together and only about 8–10 percent of drugs entering Phase I reach market.

Q: How do biotech Phase I II III probabilities affect investment decisions? Phase PoS is the central variable in risk-adjusted NPV (rNPV) models. A single asset reading out Phase II positive can triple its rNPV in one day as cumulative PoS jumps from roughly 12 to 44 percent for an oncology drug. Conversely, a Phase III failure on a large-cap pharma's lead asset can wipe 20–30 percent of market cap. Portfolio-level PoS analysis determines whether a pipeline justifies a company's valuation relative to its dilution risk from future financing needs.

Q: What is a real-world example of biotech phase probability analysis? In the worked example, an oncology Phase II asset with $2 billion peak unrisked sales has cumulative remaining PoS of 12.3 percent (28% × 50% × 88%). Risk-adjusted peak sales equal $246 million. If a positive Phase II readout occurs, cumulative PoS jumps to roughly 44 percent, tripling rNPV in a single announcement, which explains why biotech stocks trade with embedded binary optionality.

Q: How can investors use biotech Phase I II III probability analysis? Always condition PoS on therapeutic area and mechanism class, not the industry average. Check whether the trial uses a biomarker-selected population and a validated target, both materially raise PoS versus a first-in-class unvalidated mechanism. Build a separate probability for financing risk alongside clinical PoS when the company has less cash than its expected time to next readout.

Q: How are Phase II and Phase III biotech probabilities different? Phase II is the valley of death because it is the first real test of biological effect, about 71 percent of Phase II candidates fail, primarily for lack of efficacy. Phase III tests the same effect in larger, more diverse populations with a statistical hurdle sufficient for regulatory filing; about 42 percent of Phase III attempts fail to reach filing, primarily from insufficient effect size relative to placebo or unexpected safety signals in the broader population.

Sources

1. Tufts Center for the Study of Drug Development. "Cost and Success Rate Studies." https://csdd.tufts.edu/
2. Biotechnology Innovation Organization (BIO). "Clinical Development Success Rates 2011-2020." https://www.bio.org/clinical-development-success-rates-2011-2020
3. U.S. Food and Drug Administration. "The Drug Development Process." https://www.fda.gov/patients/drug-development-process
4. Pfizer Inc. Annual Report on Form 10-K. SEC EDGAR. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000078003&type=10-K

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.