SOA Exam FAM is a three-and-a-half-hour computer-based test of 34 multiple-choice questions spanning ten topics, from mortality models and premium calculation to severity, frequency, aggregate loss, credibility, and short-term pricing and reserving. This guide walks through the format, 2026–2027 sittings and fees, the official topic weights, worked examples from both halves of the syllabus, and the study materials the SOA itself provides.
Format and assumed knowledge
FAM is a three-and-a-half-hour computer-based exam with 34 multiple-choice questions, and it assumes you have already covered probability, mathematical statistics, and financial mathematics.
The Fundamentals of Actuarial Mathematics exam runs three and a half hours and consists of 34 multiple-choice questions, administered as a computer-based test. Before scheduling it, confirm you have the assumed background: the probability and financial mathematics covered in Exams P and FM, plus VEE Mathematical Statistics.
You will not need to bring tables. A set of probability distribution inventories and life and decrement tables is provided during the exam through an Exhibit button, and the testing software includes a normal distribution calculator. Paper-and-pencil candidates receive hard copies instead, and bringing your own tables into the room is not allowed.
The SOA publishes a July 2026 and an October 2026 syllabus for this exam. This guide follows the October 2026 edition; check the Updates section on the exam's home page for any changes before you register.
Sources: SOA Fundamentals of Actuarial Mathematics October 2026 syllabus, format, learning outcomes, weights, and readings — October 2026 FAM syllabus; Studying for Fundamentals of Actuarial Mathematics (FAM) Exam — SOA FAM study page
Dates, registration, and fees
The October 2026 sitting runs October 23–29 with registration closing September 22, 2026; the 2026 fee is $408.
At the time of this update, the next FAM administration is October 23–29, 2026 by CBT, with a paper-and-pencil option on October 23. Registration for that sitting closes September 22, 2026 at 11:59 PM — only days after this guide was last checked, so verify the deadline before planning around it. The following sittings are February 18–24, 2027 (registration closes January 19, 2027) and July 13–19, 2027 (registration closes June 15, 2027).
Most SOA exams are offered by computer-based testing through Prometric test centers. If you are an international candidate without access to a Prometric center, selected exams — including FAM — are available in paper-and-pencil format on the first day of the window.
The FAM exam fee is $408 for 2026, and the fee includes electronic access to the required study notes. All amounts are in U.S. dollars and fees are subject to change; reduced fees, reimbursement, and scholarship opportunities exist for eligible candidates.
Sources: Exam Schedule — SOA exam schedule, ASA and CERA listings; Exam and e-Learning Module Fees — ASA Exams and Modules 2026, updated December 16, 2025
The ten tested topics and their weights
The syllabus lists ten topics with published weight ranges; premium and policy value calculation for long-term coverages carries the largest weight at 15–22.5%.
The table below is the official October 2026 topic map. Two things to keep in mind as you plan: the weights are ranges intended to apply to most but not every individual exam, and a single question can draw on multiple learning outcomes. Treat the larger-weight topics as your study anchors, not as a guarantee of question counts.
| Topic | Weight |
|---|---|
| 1. Short-term insurance and reinsurance coverages | 5–10% |
| 2. Severity, frequency, and aggregate models | 12.5–17.5% |
| 3. Parametric estimation | 2.5–7.5% |
| 4. Introduction to credibility | 2.5–5% |
| 5. Pricing and reserving for short-term insurance coverages | 10–15% |
| 6. Option pricing fundamentals | 2.5–7.5% |
| 7. Long-term insurance coverages and retirement financial security programs | 2.5–5% |
| 8. Mortality models | 10–15% |
| 9. Present value random variables for long-term insurance coverages | 12.5–20% |
| 10. Premium and policy value calculation for long-term insurance coverages | 15–22.5% |
Two modeling families, one exam
FAM combines contingent payment models built on survival probabilities and discounted cash flows with short-term loss models built on claim frequency, severity, and estimation. Keeping the two vocabularies separate is the organizing skill.
The long-term half — Topics 7 through 10 — works with present values of payments that depend on survival. Insurances, annuities, benefit premiums, and policy values are all present values, or differences of present values, evaluated under an interest rate and a mortality assumption. Precision about when cash flows occur is the core discipline here. Topic 7 itself is definitional: insurable interest, the main long-term coverage types, annuities, and defined benefit versus defined contribution pension plans.
The short-term half — Topics 1 through 6 — works with distributions of claim counts and claim sizes. Severity models, frequency models, the aggregate loss variable, risk measures, maximum likelihood estimation, credibility, and the pricing and reserving methods form a chain: each link describes one aspect of a portfolio's loss experience. The recurring questions are which moments you can compute, which distribution fits, and how observations adjust a rate or reserve.
Topic 1 also covers practical short-term mechanics that are easy to overlook: coverage modifications, the loss elimination ratio, the effect of inflation on losses, and how proportional and excess-of-loss reinsurance split claim amounts between insurer and reinsurer.
Because similar-looking symbols carry different meanings across the two halves, keep a separate notation sheet for each. The SOA's own notation and terminology note for FAM, listed under Other Resources on the syllabus, is a useful baseline for both.
Premium and policy value questions reward precise timing
The equivalence-principle premium equates the actuarial present value of benefits with that of premiums, and policy values are the prospective differences between them. Payment modes and benefit timing determine which factors you use.
For long-term policies, the syllabus expects you to identify the future loss random variables for whole life, term life, and endowment insurance and for annuities, then calculate premiums under the equivalence principle, the portfolio percentile principle, or a target expected present value of profit. Policy values come in gross, net, and modified net varieties, and you should be able to calculate the effect of changed mortality or interest assumptions. Extra risk is modeled through age rating, a constant addition to the force of mortality, or a constant multiple of the mortality rate.
A habit worth building here is tagging before computing: write down when the benefit is paid (at the moment of death, or at the end of the year of death) and when premiums are paid (continuously, annually, or 1/m-thly). Those tags select the annuity and insurance factors. Where fractional years are involved, the syllabus expects standard fractional age or claims acceleration approximations.
Try this exercise before reading the solution.
Practice exercise
A two-year term insurance pays 1,000 at the end of the year of death. The probability of death in year 1 is 0.10 and in year 2 (conditional on survival to year 2) is 0.20. Interest is 5% effective. Premiums P are payable at the start of each year, with the second premium paid only if the insured survives. Calculate the equivalence-principle premium.
Show answer
P ≈ 139.19
Present value of benefits: death in year 1 contributes 1,000 × 0.10 / 1.05 = 95.24. Death in year 2 contributes 1,000 × 0.9 × 0.20 / 1.05² = 163.27. Total APV of benefits = 258.50. Present value of premiums: P × (1 + 0.9 / 1.05) = 1.857143 P. Setting them equal gives P = 258.50 / 1.857143 ≈ 139.19. Check your own work against the tags: end-of-year-of-death benefit and annual advance premiums — mixing either convention changes the factors and the answer.
Frequency, severity, and the aggregate loss variable
The aggregate loss S is the sum of a random number N of claim amounts X. Its mean and variance follow from two composition formulas, with a shortcut when N is Poisson.
The collective risk relationships — E[S] = E[N]·E[X] and Var(S) = E[N]·Var(X) + Var(N)·(E[X])² — are the backbone of this half. When N is Poisson with parameter λ, the variance collapses to λ·E[X²], a shortcut worth deriving once so you know exactly when it applies. Around these moments, the syllabus also expects the normal or log-normal approximation to the aggregate distribution, convolution calculations, expected stop-loss payments, and Value at Risk and Tail Value at Risk with their properties.
The other recurring distinction is between adjustment channels: inflation and coverage modifications act on severity, while exposure changes act on frequency. Confusing those two is the mix-up to guard against when a question scales a model up or down.
Work the next exercise both ways — general formula first, Poisson shortcut second — and confirm the answers agree.
Practice exercise
Claim counts N follow a Poisson distribution with mean 0.4 per period. Individual claim amounts X are exponential with mean 200. Let S be total claims in a period. Calculate E[S] and Var(S).
Show answer
E[S] = 80 and Var(S) = 32,000
E[S] = E[N]·E[X] = 0.4 × 200 = 80. For the variance, first use the general two-term formula: Var(X) = 200² = 40,000 for the exponential, so Var(S) = 0.4 × 40,000 + 0.4 × 200² = 16,000 + 16,000 = 32,000. Then verify with the compound-Poisson identity: E[X²] = 2 × 200² = 80,000, so Var(S) = λ·E[X²] = 0.4 × 80,000 = 32,000. The two routes must agree exactly; if they do not, the slip is algebraic, not conceptual. Notice both terms contribute equally here because the Poisson has Var(N) = E[N] — switching N to a negative binomial would change only the Var(N) term.
Credibility, estimation, and reserving as tested
Under the October 2026 syllabus, the credibility topic covers the concept and limited fluctuation calculations; parametric estimation covers maximum likelihood for three data situations; and pricing and reserving covers three claim-estimation methods plus ratemaking adjustments.
The credibility topic is deliberately narrow: understand the concept of credibility and perform calculations using limited fluctuation, or classical, credibility, which sets a full-credibility standard from a stability tolerance and grants partial credibility in proportion to data volume. If a study manual spends chapters on other credibility frameworks, check them against the current syllabus reading assignments before investing time.
Parametric estimation expects you to fit severity and frequency distributions by maximum likelihood for three data situations: complete individual data, complete grouped data, and truncated or censored data. Being able to tell these data situations apart — and set up the corresponding likelihood contributions — is the tested skill.
For pricing and reserving, know three methods for estimating outstanding claims: Expected Loss Ratio, Chain-Ladder, and Bornhuetter-Ferguson. As an editorial reading of the methods, chain ladder projects future losses by extrapolating the observed reporting pattern, so it is most dependable for accident years with enough development; Bornhuetter-Ferguson loads only the unreported fraction onto an a priori loss expectation, which anchors an immature year to pricing expectations instead of noisy early data. On the ratemaking side, the syllabus adds objectives and data, adjustments for development, trend, and current rate levels, expense and profit loadings, and overall rate changes by the loss cost and loss ratio methods.
Do not forget Topics 3 and 6 sit in this half too: option pricing fundamentals expect binomial and Black-Scholes pricing of simple European derivatives on a single non-dividend-paying asset, delta hedging, and put-call parity.
Study materials the SOA provides
Three textbooks with specified chapters, the official FAM tables and notation note, and two sets of sample questions with solutions form the core materials — all listed on the syllabus itself.
The required readings are Loss Models: From Data to Decisions (Fifth Edition, Klugman, Panjer, and Willmot) for the short-term topics, Introduction to Ratemaking and Loss Reserving for Property and Casualty Insurance (Fifth Edition, Brown and Lennox), and Actuarial Mathematics for Life Contingent Risks (Third Edition, Dickson, Hardy, and Waters) for the long-term topics. Each is assigned by specific chapters and sections on the syllabus, and for the life contingencies text the exercises count as required reading. Note that the texts and study notes are not available with the exam booklet — only the tables are.
Under Other Resources, the SOA provides the Tables for FAM (with a long-term Excel workbook for educational use that will not be available at the exam), the notation and terminology note, and two sample question sets with solutions: one for Topics 1–6 and one for Topics 7–10. The SOA's FAM study page also links both current syllabus editions and its study methods suggestions.
As a suggestion rather than a requirement: work the official sample questions topic by topic first, keeping the two halves separate while you build definitions, then move to mixed timed sets so that identifying which modeling family a problem belongs to becomes automatic. Failed sample questions point directly at the syllabus outcome to re-read.
Sources: SOA Fundamentals of Actuarial Mathematics October 2026 syllabus, format, learning outcomes, weights, and readings — October 2026 FAM syllabus; Studying for Fundamentals of Actuarial Mathematics (FAM) Exam — SOA FAM study page
How scoring works on exam day
Each question has five options with exactly one correct answer, unanswered questions score zero, and a few unscored pilot questions are mixed in. Unofficial pass/fail results arrive by email within an hour.
Every multiple-choice problem offers five answer choices, A through E, with exactly one correct; some answers are rounded, so read the tolerance instructions in each question. All unanswered questions are scored incorrect, which means you should answer every question — there is no penalty for a wrong guess. There is also no set requirement for how often each letter appears as the correct answer, so treat each question independently.
A few pilot questions are randomly placed in every form, in both CBT and paper formats. They are being tested for future use and do not count toward your score; all other questions do.
Because the CBT exam runs over several days, candidates receive different test forms drawn from a question pool. The SOA uses statistical scaling to keep forms comparable in content and passing criteria within the same testing period. You will receive unofficial pass/fail results by email, at the address you used to schedule your appointment, within about an hour of finishing.
Sources
Facts checked: