In 1997, Christopher Lipinski and his colleagues at Pfizer published one of the most influential papers in modern medicinal chemistry. It distilled decades of empirical knowledge about orally absorbed drugs into four simple cutoffs that have shaped how chemists approach lead optimization ever since. Today, nearly thirty years later, the “Rule of Five” (Ro5) is still printed on whiteboards in every drug discovery lab in the world.
This guide unpacks what the four rules actually say, why they work, where they break down, and how to compute them in practice. You can calculate Ro5 properties for any molecule in the Drug Discovery Lab or with the free Molecular Weight Calculator.
The four rules
Lipinski analyzed the World Drug Index — about 2,200 compounds that had progressed to phase II trials with reasonable oral bioavailability — and asked which physicochemical properties these drugs had in common. Four cutoffs emerged, each at either 5 or a multiple of 5:
- Molecular weight ≤ 500 Da. Above 500, passive permeability across cell membranes drops sharply.
- cLogP ≤ 5. The calculated octanol-water partition coefficient measures lipophilicity. Above 5, the compound is too greasy and tends to have poor solubility, high promiscuity, and poor selectivity.
- Hydrogen bond donors ≤ 5. Counted as the number of N-H and O-H bonds in the molecule. Donors stick to water and other polar surfaces, slowing membrane crossing.
- Hydrogen bond acceptors ≤ 10. Counted as the number of N and O atoms (Lipinski's simple approximation). Acceptors also pay an energetic cost to desolvate before crossing a membrane.
Lipinski's formal “rule” is that a compound is likely to have poor absorption or permeation if it violates more than one of these cutoffs. One violation is considered acceptable.
Why these cutoffs work
The Rule of Five works because it captures the basic physics of passive intestinal absorption in a few easy-to-compute numbers. Drug absorption from the gut is dominated by two processes: dissolution in intestinal fluid and crossing the apical membrane of enterocytes.
- Solubility in the aqueous intestinal lumen drops as molecular weight rises and as lipophilicity increases. Big greasy molecules are hard to dissolve.
- Permeability across the lipid bilayer requires the molecule to shed its solvation shell and pass through a hydrophobic environment. Each polar group has to break its hydrogen bonds with water before it can cross, which is energetically costly. Too many donors or acceptors means too much desolvation penalty.
These two trends pull in opposite directions: more polarity helps solubility but hurts permeability, and more lipophilicity helps permeability but hurts solubility. Ro5 is essentially a goldilocks zone where both can be reasonable simultaneously. Compounds that fall in that zone have a much better chance of being absorbed orally without aggressive formulation work.
Calculate Ro5 in SciRouter
The Drug Discovery Lab calculates Ro5 properties on demand. Here is an example using the API directly:
curl -X POST https://scirouter.ai/v1/drug-discovery-lab/calculate/molecular-properties \
-H "Authorization: Bearer sk-sci-..." \
-H "Content-Type: application/json" \
-d '{
"smiles": "CC(=O)Oc1ccccc1C(=O)O"
}'The response includes molecular weight, cLogP, hydrogen bond donors, hydrogen bond acceptors, TPSA, rotatable bonds, ring count, and a count of Ro5 violations. The example SMILES above is aspirin, which (unsurprisingly) passes all four rules with room to spare.
Beyond Lipinski: where Ro5 doesn't apply
Lipinski himself was always clear that Ro5 was a guideline for passively absorbed oral drugs targeting traditional binding pockets. Several modern drug classes break the rules intentionally — and successfully:
Kinase inhibitors
Many approved kinase inhibitors push or break Ro5. Imatinib (493 Da, cLogP 3.0, 2 donors, 7 acceptors) just barely fits. Sorafenib, Nilotinib, and several others are heavier and greasier. The kinase ATP pocket favors larger, more lipophilic scaffolds and the field accepted the Ro5 violations because the alternatives didn't work.
PROTACs and molecular glues
Proteolysis-targeting chimeras (PROTACs) intentionally break Ro5. They are bifunctional molecules — one ligand for the target protein, a linker, and another ligand for an E3 ubiquitin ligase — and routinely have molecular weights of 800-1200 Da, cLogP above 5, and many H-bond donors/acceptors. Despite that, several PROTACs have shown oral bioavailability in humans, which has driven the field to develop new design rules specifically for the “PROTAC chemical space.”
Macrocycles
Cyclic peptides and macrocyclic natural products like cyclosporine, tacrolimus, and rapamycin are well over 500 Da with many H-bond donors and acceptors. Yet some are orally bioavailable. The trick is that macrocycles can adopt conformations that internally satisfy their own H-bonds, effectively shielding the polar groups from solvent and reducing the desolvation penalty when they cross a membrane. This phenomenon — sometimes called “chameleonic behavior” — has become a major design strategy for beyond-Ro5 chemistry.
Peptides and biologics
Therapeutic peptides like Semaglutide and Insulin are far beyond Ro5 by every measure. They are not passively absorbed orally — they require subcutaneous injection or specialized absorption-enhancer formulations — and that's exactly the point. Once you decide a different route of administration is acceptable, Ro5 stops applying.
QED and other improvements
The Rule of Five is a hard four-cutoff filter, which has the downside that a compound at MW 499 is “passing” and at MW 501 is “failing” even though they're essentially identical. To smooth this out, Bickerton and Hopkins introduced the QED (Quantitative Estimate of Drug-likeness) score in 2012.
QED combines eight properties — molecular weight, cLogP, H-bond donors, H-bond acceptors, polar surface area, rotatable bonds, aromatic rings, and a count of structural alerts — into a single 0-1 score using desirability functions. Each property gets a smooth curve from 0 to 1 calibrated to the distribution of approved drugs, and the eight scores are combined as a geometric mean. A QED of 0.9 is excellent, 0.5 is borderline, 0.2 is very poor.
Other related metrics include Veber's rules (rotatable bonds and PSA), the Egan egg, and Ghose's filter — all of which encode similar physicochemical intuition with slightly different cutoffs. Most modern medicinal chemistry tools compute several of these in parallel.
Practical advice for using Ro5 well
- Use it as a soft signal, not a binary filter.A compound with one violation but interesting biological activity is still a lead worth pursuing.
- Track lipophilicity (cLogP) more closely than the other properties. High cLogP correlates with attrition in clinical development more strongly than molecular weight or H-bond counts alone.
- Calculate ligand efficiency. A 200 Da fragment with micromolar affinity has more headroom for optimization than a 600 Da Ro5-violator with the same affinity.
- Don't apply Ro5 to chemistry it wasn't designed for. Peptides, macrocycles, and PROTACs need their own design rules.
Explore in the Drug Discovery Lab
The molecule editor in the Drug Discovery Lab lets you sketch or paste a SMILES, compute Lipinski properties on the fly, and instantly see which rules pass or fail. You can also load any existing drug workspace (Aspirin, Imatinib, Semaglutide, and more) and compare its Ro5 profile against your own design.
Or use the free Molecular Weight Calculator for a no-login version that runs the same RDKit-backed calculation on a single SMILES at a time.
Bottom line
Lipinski's Rule of Five works because it summarizes the basic biophysics of intestinal absorption in four numbers any chemist can compute in seconds. It is not a hard rule and modern drug discovery routinely produces successful compounds that violate it — but those exceptions tend to require explicit engineering to overcome the absorption challenges Ro5 was designed to flag. Used as a guideline, not a filter, the Rule of Five remains one of the most productive heuristics in all of medicinal chemistry.