BEFORE THE METHOD TABLES
Two jobs, not one
Preventing pregnancy and reducing infection are separate jobs, and a method can be good at one while doing nothing at all for the other. Reading a figure from one job as though it answered the other is the mistake this page exists to prevent.
So decide which question you are asking before you read a number. Everything below is sorted by the job it is published to do.
Where the numbers on this page come from
Every method here is carried from one reviewed research dataset, and every published figure is reproduced exactly as that dataset records it, including the wording in brackets that says who published it. Nothing on this page is rounded, averaged, converted, or recalculated.
Every method in that dataset appears below. None has been left out to simplify the picture, and the order within each group is the dataset's own. It is not a ranking, and nothing here recommends one method over another.
What this page is not
This is reference material, not medical advice, and clinical review of it has not happened yet. It does not calculate anyone's personal risk, and it carries no per-act infection figures. Those limits are described at the end, and they are deliberate.
How to read these figures
Published effectiveness figures are easy to misread. The rules below are the ones that matter most for the tables that follow.
Typical use and perfect use answer different questions
Typical use describes what happened to people using a method in ordinary life, including the occasions it was used late, used wrongly, or not used. Perfect use describes the method followed exactly as directed every single time. Each is published as pregnancies per 100 users over the time span named in the table.
A pair of figures usually comes from two different bodies
For every method below that carries a figure in each column, those two figures were not published by the same organisation. That pairing is therefore not one study's spread between careful and careless use, and the tables label it: “Typical and perfect figures come from different sources”. Each figure names its own source in brackets, so you can always see which body stands behind which number.
A missing figure is not a zero
Where the reviewed sources published no figure, the tables say “No published figure”. That is not zero, it is not “completely effective”, and it is not comparable to a method that has a number. Several methods below have a typical-use figure and no perfect-use figure. Where the sources gave a reason it appears in that method's cautions; where they did not, the gap has been left as a gap rather than filled in.
One estimate can be shared across methods
The diaphragm and the cervical cap share a single published perfect-use estimate even though their typical-use figures differ. Wherever one estimate is pooled like that, the table names every method it is pooled across, so it can never read as separate measurements that happen to agree.
Competing estimates are shown, not reconciled
Some methods have more than one published typical-use estimate. Where that happens the competing figure is shown alongside the first, with a note on why the two are not merged. Averaging estimates drawn from different study pools would produce a number that no source published.
None of these can be combined into a personal risk
These are population estimates. They cannot be multiplied, added, or otherwise combined into an individual probability, and this page does not offer a way to try. Where a figure is a range, the range is there to be read as a range.
Barriers
Barriers are the only group on this page with an established role in reducing infection, which is why each one is described in full rather than tabulated.
Only the external and the internal condom carry published pregnancy figures. For every other barrier here, the absence of a figure is a statement about the published evidence rather than a verdict on the method.
None of the statements below is a percentage, and that is deliberate: per-act, infection-specific estimates are not published in this dataset. Each infection's own page carries the sourced transmission panel instead.
External condom (latex)
Infection: Reduces acquisition of fluid-transmitted infections when used throughout contact. Protection is incomplete for infectious skin outside the covered area; do not apply one reduction to every STI.
Using it: Use a new intact condom for each act and change between partners or anatomical sites. Use compatible lubricant and hold the base during withdrawal.
Cautions: Do not combine with another external or internal condom. Latex allergy requires a suitable alternative. Pregnancy estimates are annual, not per-act STI estimates.
Pregnancy: Typical use 13 (CDC US). Perfect use 2 (NHS UK). Measured over: First year of use unless otherwise stated. Typical and perfect figures come from different sources.
Also published: 18 per 100 in year one — NHS estimate is not averaged with CDC. 15 per 100 per year — NHS inform uses another estimate; no harmonization attempted.
Evidence confidence: Medium. Clinical review pending.
Synthetic external condoms (polyurethane/polyisoprene)
Infection: An STI-prevention option when an appropriate regulated product is used. Material- and product-specific failure rates are not interchangeable.
Using it: Check the product instructions and lubricant compatibility; polyisoprene is not compatible with oil-based lubricant.
Cautions: The general external-condom pregnancy figure is not a separate validated estimate for each material.
Pregnancy: No published figure. The dataset records this method's pregnancy unit as “not quantified / not applicable”.
Evidence confidence: Medium. Clinical review pending.
Natural-membrane condom
Infection: Not a dependable choice for HIV or other STI prevention.
Using it: Do not substitute for a latex or suitable synthetic barrier when STI prevention is a goal.
Cautions: Pregnancy prevention does not establish pathogen protection.
Pregnancy: No published figure. The dataset records this method's pregnancy unit as “not quantified / not applicable”.
Evidence confidence: Medium. Clinical review pending.
Internal condom (also called female condom)
Infection: May reduce STI transmission; evidence is less extensive than for external condoms, and vaginal evidence cannot be automatically transferred to anal use.
Using it: Follow product instructions and ensure penetration enters the pouch, not the space beside it.
Cautions: Do not use simultaneously with an external condom. Annual pregnancy rates concern vaginal use; no independent per-act STI reduction is asserted.
Pregnancy: Typical use 21 (CDC US and NHS UK). Perfect use 5 (NHS UK). Measured over: First year of use unless otherwise stated. Typical and perfect figures come from different sources.
Evidence confidence: Medium. Clinical review pending.
Dental dam / appropriate cut-open condom
Infection: Limits mouth-to-vulva or mouth-to-anus contact over the covered area. No reliable organism-specific clinical percent reduction is established in the reviewed guidance.
Using it: Place before contact; retain the same orientation; use a fresh barrier for each contact. Use compatible lubricant.
Cautions: Not contraception for vaginal intercourse. Not protection for separate kissing or uncovered lesions.
Pregnancy: No published figure. The dataset records this method's pregnancy unit as “not quantified / not applicable”.
Evidence confidence: Low. Clinical review pending.
Gloves and finger cots
Infection: Contact-barrier rationale for fingers and small skin breaks; no validated per-act STI-effectiveness percentage supplied.
Using it: Use intact barriers and replace between people and body sites; avoid moving contaminated lubricant or material to a clean site.
Cautions: Do not advertise as tested pregnancy contraception or as eliminating transmission. Practical extrapolation needs clinical/editorial review.
Pregnancy: No published figure. The dataset records this method's pregnancy unit as “not quantified / not applicable”.
Evidence confidence: Low. Clinical review pending.
Sex-toy barriers and cleaning
Infection: Reduces transfer opportunities when barriers are changed and toys cleaned appropriately; a universal effectiveness percentage is unavailable.
Using it: Use a new compatible condom on toys where appropriate and follow material/manufacturer cleaning instructions.
Cautions: Porous materials and devices with seams can be difficult to clean. Cleaning is not a validated substitute for sterilization of equipment that punctures skin.
Pregnancy: No published figure. The dataset records this method's pregnancy unit as “not quantified / not applicable”.
Evidence confidence: Low. Clinical review pending.
Supporting measures
A supporting measure helps a barrier work. It is not itself a barrier, a contraceptive, or an infection-prevention method, and the dataset keeps it in its own group so it cannot be read as one.
Compatible lubricant
Infection: Helps use a barrier comfortably; not an antimicrobial, contraceptive or independently quantified STI-prevention method.
Using it: Use lubricant compatible with the barrier and toy material.
Cautions: Oil damages latex and polyisoprene. Irritation is a reason to stop and reassess the product.
Pregnancy: No published figure. The dataset records this method's pregnancy unit as “not quantified / not applicable”.
Evidence confidence: Medium. Clinical review pending.
Contraception
These methods are published for pregnancy prevention.
On infection the dataset records the same statement for every method here: “No established STI protection.”
That is why a barrier is a separate decision rather than an upgrade to one of these.
Figures are pregnancies per 100 users over the span in the final column. Read the two figure columns as answers to different questions, not as a range.
| Method | Typical use | Perfect use | Measured over |
|---|---|---|---|
Copper IUD | 0.8 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Levonorgestrel IUD / IUS | 0.1-0.4 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Contraceptive implant | 0.1 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Progestin injection | 4 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Combined oral contraceptive | 7 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Progestin-only pill | 7 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Contraceptive patch | 7 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Vaginal contraceptive ring | 7 (CDC US) | Less than 1 (NHS UK) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Diaphragm with spermicide | 17 (CDC US) | 4-8 (NHS combined diaphragm/cap estimate) One estimate shared across methods: Cervical cap with spermicide Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Cervical cap with spermicide | 22 (CDC US) | 4-8 (NHS combined diaphragm/cap estimate) One estimate shared across methods: Diaphragm with spermicide Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Contraceptive sponge | 17 (CDC US) | No published figure | First year of use unless otherwise stated |
Spermicide / vaginal pH-regulating products | 21 (CDC overview) | No published figure | First year of use unless otherwise stated |
Vasectomy | 0.15 (CDC US) | No published figure | First year of use unless otherwise stated |
Tubal surgery | 0.5 (CDC US) | No published figure | First year of use unless otherwise stated |
Contraception in practice
The same methods, with what the sources say about using them and what they warn about. Evidence confidence is the dataset's own grade for that record, and every record on this page is awaiting clinical review.
| Method | How it is used | Cautions | Evidence confidence |
|---|---|---|---|
Copper IUD | Clinician placement; ongoing hormone-free contraception. | Discuss bleeding/pain, contraindications and product-specific duration. | Medium |
Levonorgestrel IUD / IUS | Clinician placement; product-specific dose and duration. | Bleeding-pattern changes and placement considerations. | Medium |
Contraceptive implant | Clinician placement and removal. | Bleeding changes and interacting medications; check current product duration. | Medium |
Progestin injection | Receive on the prescribed schedule. | Discuss bleeding changes, bone health and timing of return to fertility. Also published: 6 per 100 in year one — Different published typical-use estimate. | Medium |
Combined oral contraceptive | Take according to the specific formulation; use its missed-pill instructions. | Estrogen-related contraindications and interacting drugs require review. Also published: 9 per 100 in year one — Different published typical-use estimate. | Medium |
Progestin-only pill | Formulation-specific timing and missed-pill rules. | Do not use a single missed-pill rule for all products. Also published: 9 per 100 in year one — Different published typical-use estimate. | Medium |
Contraceptive patch | Replace according to product schedule. | Estrogen eligibility, detachment and formulation-specific weight cautions. Also published: 9 per 100 in year one — Different published typical-use estimate. | Medium |
Vaginal contraceptive ring | Follow formulation-specific insertion/removal schedule. | Estrogen eligibility and time outside the vagina depend on product. Also published: 9 per 100 in year one — Different published typical-use estimate. | Medium |
Diaphragm with spermicide | Fitting/instruction and product-specific spermicide and removal rules. | NHS does not provide a typical-use figure here; no STI protection. | Medium |
Cervical cap with spermicide | Use with spermicide per instructions. | Effectiveness varies with population and childbirth history. | Medium |
Contraceptive sponge | Follow placement and removal instructions. | Not STI prevention; childbirth history can affect effectiveness. | Medium |
Spermicide / vaginal pH-regulating products | Use the exact product directions. | Different products should not be treated as clinically equivalent; nonoxynol-9 irritation is relevant to HIV/STI prevention. | Medium |
Vasectomy | Use backup until post-vasectomy semen assessment clears the procedure. | Not immediately effective; intended to be permanent. | Medium |
Tubal surgery | Specialist counseling and procedure-specific information. | Intended to be permanent; technique affects outcomes. | Medium |
Behavioral methods
These depend on timing and on what people do rather than on a device or a medicine, so most of them carry no published failure figure at all. Where a figure exists it covers a family of methods that are not interchangeable with each other.
On infection the dataset does not record one statement for this whole group, so each statement appears below with the methods it covers.
Fertility-awareness methods (method-specific, not one technique), Calendar/rhythm tracking and prediction-only apps, Instructor-taught symptothermal method, Cervical-mucus fertility-awareness methods, Withdrawal, Lactational amenorrhea method (LAM): No established STI protection.
Avoiding sperm exposure to the vagina: STI transmission can still occur through oral, anal, skin contact or kissing, depending on infection.
| Method | Typical use | Perfect use | Measured over |
|---|---|---|---|
Fertility-awareness methods (method-specific, not one technique) | 2-23 (CDC range across methods) | 1-9 (NHS broad estimate) Typical and perfect figures come from different sources | First year of use unless otherwise stated |
Calendar/rhythm tracking and prediction-only apps | No published figure | No published figure | First year of use unless otherwise stated |
Instructor-taught symptothermal method | No published figure | No published figure | First year of use unless otherwise stated |
Cervical-mucus fertility-awareness methods | No published figure | No published figure | First year of use unless otherwise stated |
Withdrawal | No published figure | No published figure | First year of use unless otherwise stated |
Lactational amenorrhea method (LAM) | No published figure | No published figure | First six months postpartum, only while all eligibility conditions hold |
Avoiding sperm exposure to the vagina | No published figure | No published figure | First year of use unless otherwise stated |
| Method | How it is used | Cautions | Evidence confidence |
|---|---|---|---|
Fertility-awareness methods (method-specific, not one technique) | Learn a defined method; avoid intercourse with pregnancy potential or use a barrier during the method-defined fertile days. | These broad ranges are not a head-to-head ranking. App forecasts, mucus methods, temperature-based methods and trained symptothermal methods are not interchangeable. Also published: 24 per 100 in year one — Broad NHS natural-family-planning figure, not the same study pool. | Medium |
Calendar/rhythm tracking and prediction-only apps | Calendar prediction uses previous cycles; it does not confirm the current cycle's ovulation. | Do not borrow efficacy from a separately studied algorithm or instructor-taught method. Irregular cycles complicate prediction. | Low |
Instructor-taught symptothermal method | Combines temperature with other signs under a named protocol, rather than guessing a single fertile day. | Study participants, training, abstinence rules and adherence affect results. Named-method figures are held for full-text/correction review. | Medium |
Cervical-mucus fertility-awareness methods | Use a defined protocol and trained interpretation of mucus changes. | Not equivalent to casual observation or an LH-only contraceptive strategy; no STI protection. | Medium |
Withdrawal | Requires withdrawal before ejaculation and avoiding semen at the vaginal opening. | No numerical failure estimate added without a verified source table for this method; not STI protection or emergency contraception. | Low |
Lactational amenorrhea method (LAM) | Requires no return of menses, fully or nearly fully breastfeeding, and less than six months since birth. | Switch methods when any condition stops being met; the observation window is not one year. | Medium |
Avoiding sperm exposure to the vagina | Describe the actual activities rather than labeling all nonpenetrative sex safe. | No annual pregnancy failure rate assigned to an undefined collection of activities; semen near the vaginal opening matters. | Low |
Emergency contraception
Emergency contraception is measured per episode, not per year, so it has no annual failure figure to compare against the methods above. Comparing it to them on a single number would be comparing different things.
On infection the dataset records the same statement for every method here: “No established STI protection.”
| Method | Typical use | Perfect use | Measured over |
|---|---|---|---|
Levonorgestrel emergency contraceptive pill | No published figure | No published figure | One emergency-contraception episode; not an annual or per-act failure estimate |
Ulipristal emergency contraceptive pill | No published figure | No published figure | One emergency-contraception episode; not an annual or per-act failure estimate |
Copper IUD emergency contraception | No published figure | No published figure | One emergency-contraception episode; not an annual or per-act failure estimate |
52-mg levonorgestrel IUD for emergency contraception | No published figure | No published figure | One episode; quantitative review outstanding |
| Method | How it is used | Cautions | Evidence confidence |
|---|---|---|---|
Levonorgestrel emergency contraceptive pill | Seek care promptly after contraceptive failure or intercourse with pregnancy potential; options are generally considered within five days, with method-specific criteria. | No STI protection and not protection for later intercourse. After ulipristal, timing of hormonal-method restart needs specific advice; do not interchange with levonorgestrel instructions. | Medium |
Ulipristal emergency contraceptive pill | Seek care promptly after contraceptive failure or intercourse with pregnancy potential; options are generally considered within five days, with method-specific criteria. | No STI protection and not protection for later intercourse. After ulipristal, timing of hormonal-method restart needs specific advice; do not interchange with levonorgestrel instructions. | Medium |
Copper IUD emergency contraception | Seek care promptly after contraceptive failure or intercourse with pregnancy potential; options are generally considered within five days, with method-specific criteria. | No STI protection and not protection for later intercourse. After ulipristal, timing of hormonal-method restart needs specific advice; do not interchange with levonorgestrel instructions. | Medium |
52-mg levonorgestrel IUD for emergency contraception | Discuss with a clinician using local guidance. | Evidence and jurisdiction-specific adoption require additional review; not listed as an EC option in the retrieved CDC page. No equivalent-to-copper claim published here. | Low |
Using more than one method
Because pregnancy and infection are separate jobs, people often use a barrier alongside contraception. Each method then keeps its own published figure: the contraception carries the pregnancy figure, the barrier carries the infection role.
The two figures are not merged into one number on this page. No source publishes a combined figure, and producing one by arithmetic would invent a result that no study measured.
Two barriers at once
External condom (latex): Do not combine with another external or internal condom. Latex allergy requires a suitable alternative. Pregnancy estimates are annual, not per-act STI estimates.
Internal condom (also called female condom): Do not use simultaneously with an external condom. Annual pregnancy rates concern vaginal use; no independent per-act STI reduction is asserted.
Compatibility is a safety question
Compatible lubricant: Oil damages latex and polyisoprene. Irritation is a reason to stop and reassess the product.
Spermicide / vaginal pH-regulating products: Different products should not be treated as clinically equivalent; nonoxynol-9 irritation is relevant to HIV/STI prevention.
Limits, evidence and review
No per-act infection figures
This page publishes no per-act, infection-specific risk numbers, because the dataset behind it holds none. For the barriers, the reviewed guidance points that question at the transmission evidence instead, which is where each infection's own sourced panel comes from.
Infection-specific material lives on the topic pages, and the practical testing guidance is collected under testing and prevention. How to frame the question in the first place is covered in use the right question, and the modelling behind the topic pages is in evidence and models.
No personal risk calculator
There is no tool here that turns these figures into an individual probability, and that is a deliberate limit rather than a missing feature. The research these numbers come from declined to combine its own estimates that way, and a page that did it anyway would be presenting arithmetic as evidence.
Clinical review is pending
No clinician has signed off on this page. Every record behind it carries the status “Clinical review pending”, and that remains true until an independent review happens.
Evidence date
The research behind these figures was last checked on 2026-10-27. Guidance and published estimates change, so a figure here is current as of that date and not necessarily today.
Terms used across these pages are defined in the glossary, and telling partners about a positive result is covered in partner notification.
If you want to act on this, the suggested next steps after protection and contraception are listed separately.