The data and research behind Fertility Forecaster — including how we estimate your chances of getting pregnant at every age, whether egg freezing is worth it, and when to start trying for a baby.
Fertility Forecaster is a free, evidence-based calculator that estimates your probability of having the family you want. It answers questions like: When should I start trying to conceive? What is the cost of waiting 1 or 2 more years? How much does IVF improve my chances? Should I freeze my eggs? The model uses published medical research to simulate 10,000 couples and show you how age, lifestyle, egg freezing, and IVF affect your fertility outcomes.
At a high level, the simulation runs a loop for 10,000 virtual couples. For each virtual couple:
After simulating all 10,000 couples, the model reports what percentage achieved the desired family size.
The starting per-cycle chance of getting pregnant naturally, before any adjustments for age, BMI, or smoking. We set this at 23% for the youngest women in the model (ages 18–24). From there, it declines with age.
Your per-cycle odds of getting pregnant decline with age. The model uses two different curves depending on whether you've been pregnant before:
Why the difference? Women who have conceived before have demonstrated that their reproductive system can achieve pregnancy. Women who have never been pregnant include some who may have undiagnosed fertility issues (like blocked tubes or ovulation problems), which drags the group average down.
Which curve the model uses depends on whether you report a prior pregnancy when setting up the simulation. If you do, the “been pregnant before” curve is used. If you don't, the “never been pregnant” curve is used for the entire run.
Male age does not meaningfully affect conception odds. Our model only factors male age into miscarriage risk (see Section 4).
Not every pregnancy results in a live birth. Miscarriage risk varies with age:
| Age group | Miscarriage rate |
|---|---|
| < 29 | 9.8% |
| 30–34 | 10.8% |
| 35–39 | 16.7% |
| 40–44 | 32.2% |
| 45+ | 53.6% |
If you've had consecutive miscarriages, the risk for the next pregnancy increases:
| Consecutive miscarriages | Risk multiplier |
|---|---|
| 0 | 1.0× (baseline) |
| 1 | 1.5× |
| 2 | 2.2× |
| 3+ | 4.0× |
After a miscarriage in the simulation, the couple waits 3 months before trying again. The consecutive miscarriage count resets to zero after a successful live birth.
If the father is 35 or older, the risk of miscarriage increases modestly:
| Father's age | Effect on miscarriage risk |
|---|---|
| < 35 | 1.0× (baseline) |
| 35–39 | 1.15× |
| 40–44 | 1.23× |
| 45+ | 1.43× |
BMI has a modest effect on per-cycle conception odds. Below BMI 35, the impact is negligible. Above 35, it becomes meaningful:
| BMI | Effect on conception odds |
|---|---|
| < 18.5 (underweight) | 1.05× (slightly higher) |
| 18.5–24 (normal) | 1.00× (baseline) |
| 25–29 (overweight) | 1.01× (negligible) |
| 30–34 | 0.98× (negligible) |
| 35–39 | 0.78× (−22%) |
| 40–44 | 0.61× (−39%) |
| 45+ | 0.42× (−58%) |
Women with a BMI of 30 or higher have about a 15% lower chance of a live birth from IVF compared to normal-weight women. This applies to all IVF pathways (fresh IVF, frozen eggs, and frozen embryos).
Smoking reduces your per-cycle odds of getting pregnant naturally:
| Smoking status | Effect on conception odds |
|---|---|
| Never smoked | 1.00× (baseline) |
| Former smoker | 0.89× (−11%) |
| Occasional smoker | 0.88× (−12%) |
| Regular smoker (≥10/day) | 0.77× (−23%) |
This adjustment applies to natural conception only, not IVF.
Not every couple has the same fertility. The 23% average is just that — an average. Some couples are naturally more fertile, and some less so, due to factors the model can't directly measure (like ovarian reserve, sperm quality, or uterine receptivity). We use a beta distribution to model the variation.
To handle this, each simulated couple is randomly assigned a personal fertility rate at the start. That rate is fixed for life and represents their inherent “fertility type.” The chart below shows how these rates are distributed across couples. The dots mark the 5th, 25th, 50th, 75th, and 95th percentiles. The dashed line marks the 23% population mean. Hover over the curve for details.
A couple at the 95th percentile has roughly a coin-flip chance of getting pregnant each month, while a couple at the 5th percentile faces long odds. Age-related decline is then applied on top of each couple's personal rate.
If you've already been trying: If you tell the model you've been trying for several months without success, it adjusts downward — the logic being that highly fertile couples would have already conceived by now, so you're more likely to be in the lower part of the distribution:
| Months trying | Estimated average rate | Likely range (middle 90%) |
|---|---|---|
| 0 (just starting) | 23.0% | 1.4% – 59.6% |
| 6 months | 9.5% | 0.5% – 27.3% |
| 12 months | 6.0% | 0.3% – 17.6% |
| Age | Chance of live birth per IVF cycle |
|---|---|
| < 35 | 40.5% |
| 35–37 | 31.7% |
| 38–40 | 21.3% |
| 41–42 | 11.0% |
| 43–44 | 4% |
| 45+ | 1% |
Each IVF cycle takes about 4 months. This accounts for ovarian stimulation, egg retrieval, embryo culture, transfer, and the recommended rest period between cycles. By default, the model allows up to 3 fresh IVF cycles per child. The counter resets after each live birth.
Hirsch et al., 2024; Namath et al., 2025; Barrett et al., 2024
About 78.5% of frozen eggs survive the thawing process. The chance of a live birth from each surviving egg depends on how old you were when they were frozen:
| Age when eggs were frozen | Live birth rate per egg |
|---|---|
| < 35 | 13% |
| 35–37 | 9% |
| 38–40 | 6% |
| > 40 | 4% |
Eggs frozen at 30 retain the quality of a 30-year-old's eggs, even if used at 40. Up to 9 eggs are thawed per cycle.
SART 2023 Outcome Tables; Barrett et al., 2024
IVF with frozen embryos are generally more effective than using frozen eggs. Success depends on the woman's age when the embryos were created, not her age at transfer.
Embryos that have been genetically screened (PGT-A) and confirmed chromosomally normal have much higher success rates, because the main age-dependent cause of failure — chromosomal abnormalities — has been removed. The genetically screened curve is notably flat across all ages:
Fertility Forecaster's model is based on the family completion modeling approach from Habbema et al. 2015, which produced this table:
| Confidence level | 1 child | 2 children | 3 children |
|---|---|---|---|
| Without IVF | |||
| 50% | 41 | 38 | 35 |
| 75% | 37 | 34 | 31 |
| 90% | 32 | 27 | 23 |
| With IVF | |||
| 50% | 42 | 39 | 36 |
| 75% | 39 | 35 | 33 |
| 90% | 35 | 31 | 28 |
Our model uses the same core simulation approach, but uses newer data and extends it in several ways:
A key difference is how quickly natural conception rates decline with age. Our model uses newer data that shows a more gradual decline through the 30s, and uses separate curves for women who have and haven't been pregnant before:
Habbema's paper reports they use an average 23% per-cycle fertility rate from ages 20-30, and otherwise does not explain how they model their age-decline curve. I reverse-engineered their age decline curve from their published data, which shows a sharp drop to ~9% at age 31. They may use a more gentle age-decline curve from 20 to 30 that still averages out to 23%, but this should still work out to a < 10% per-cycle fertility rate past 30.
The curves converge by the early 40s but differ substantially in the 30s, where our model retains much higher per-cycle rates.
Additionally, we model the spread of base fertility rates differently. Based on reverse-engineering Habbema's model, we found they most likely use a normal distribution and account for subfertile and sterile couples by adding in a sterility curve. Our model does not use a sterility curve and instead uses a beta-genomic distribution and assigns a wider spread of individual fertility rates than Habbema's — creating more very-low-fertility couples to account for the fact that we don't model sterility directly.
Additionally, if you report a prior pregnancy, our model uses the more favorable “been pregnant before” curve. Habbema's model does not make this distinction.
Habbema used miscarriage rates from 2004. We use newer data (see Section 3) with more precise age-specific rates, along with adjustments for recurrent miscarriage and male partner age — neither of which Habbema modeled. Our miscarriage rates are more pessimistic at older ages: we use 32.2% at 40–44 and 53.6% at 45+, compared to Habbema's ~25% and ~35%.
Habbema's IVF success rates came from data available before 2015. IVF outcomes have improved meaningfully since then, particularly for older women. Our model uses SART 2023 data (see Section 9).
Habbema's model only included fresh IVF. Ours adds frozen eggs, frozen embryos, and genetically screened embryos (see Sections 10–11). You can enter your actual frozen inventory and the model will use it before falling back to fresh IVF.
Habbema's model did not account for BMI or smoking. Ours adjusts natural conception odds for both (see Sections 5–7). The effects are modest below BMI 35 and for former smokers, but substantial at higher BMI or for regular smokers.
Habbema's model hard-codes several assumptions:
In our model, all of these are user-configurable — you can set the minimum spacing between births, how many months of natural trying before IVF, and how many IVF cycles per child you're willing to attempt. This lets you model scenarios closer to your actual plans rather than relying on the study's assumptions.
This model gives you population-level estimates. Your individual situation depends on many factors it can't measure, including ovarian reserve (AMH), specific conditions like PCOS or endometriosis, sperm quality, and genetic factors.
The data comes primarily from North American and Northern European populations. Results may not apply equally across all backgrounds. Additionally, the model doesn't take into consideration:
Fertility Forecaster was built by a software engineer, not a fertility researcher or statistician. I have no professional background in reproductive medicine or statistics beyond two college-level statistics courses taken as part of a computer science degree. Mostly, I'm a mom who wished this type of information had been available when I was in my 20s and trying to figure out whether to freeze my eggs, when to start trying, and how to think about the tradeoffs.
This model has not been peer reviewed. The source code and methodology are available open source at github.com/joyceyan/fertility-forecaster so that others can inspect, critique, and improve it.