Every rate, schedule, and decision rule the model applies — written out in full, including what it deliberately leaves out. If you want to check the calculator's reasoning against your own, this is the page to read.
The RRSP & RRIF Meltdown Calculator answers one question: given a fixed spending target, which sequence of withdrawals across your accounts — and which CPP and OAS start ages — leaves the most money after all tax is paid, including the tax due on the final return?
Spending is held constant across every scenario the model tests. That matters, because it makes the comparison clean. Two plans that fund the identical lifestyle can differ by six figures in total tax, and the difference shows up as after-tax wealth remaining at the end of the plan. The model maximizes that remaining wealth.
Total tax is measured as lifetime tax plus terminal tax — the amount triggered when registered balances are deemed disposed at death. Counting only lifetime tax would flatter any plan that simply defers RRSP withdrawals, since deferral moves tax into the final return rather than eliminating it. The point of a meltdown strategy is to decide when that tax is paid and at what marginal rate, so both halves have to be on the scoreboard.
Rates and schedules reviewed August 2026 · 2026 brackets, RRIF minimums, LIF maximums, and OAS recovery threshold
Every result is a difference between two complete plans run over the same horizon, funding the same spending, on the same balances. One is the path most Canadians take by default. The other is the one the model searches for.
The unplanned path is not a strawman. It is what happens when nobody makes an active decision: registered money is the account you were told not to touch, so it sits, and the tax it accrues is invisible until the year the RRIF conversion forces the issue. The calculator's headline figure is the gap between that outcome and the best plan the model can find.
One structural note. The baseline is not permitted to run out of money either. If non-registered and TFSA balances are exhausted and spending is still unfunded, the model forces additional registered withdrawals at whatever marginal rate applies. Both plans fund the same lifestyle to the end of the horizon; they just pay different amounts of tax to do it.
The model uses combined federal and provincial marginal rate tables for 2026, applied bracket by bracket to taxable income, with thresholds indexed to inflation every year of the projection.
Ten provinces are modelled: Ontario, Alberta, British Columbia, Manitoba, New Brunswick, Newfoundland and Labrador, Nova Scotia, Prince Edward Island, Quebec, and Saskatchewan. Each is stored as an ordered set of bracket ceilings and combined marginal rates. Tax is computed by walking the brackets in order and applying each rate to the income falling within that band — not by applying a single average rate to the whole amount.
The lowest band in each province carries a zero rate, standing in for the basic personal amount. Above that, rates are the published combined federal-plus-provincial figures for 2026, rounded for legibility.
Bracket thresholds are multiplied each projection year by (1 + inflation)^years, matching the CRA's practice of indexing brackets annually. Rates themselves stay fixed; only the thresholds move.
This is not a cosmetic detail. Over a thirty-year horizon, holding thresholds at nominal 2026 levels while letting income inflate would push every plan into higher and higher brackets for no real reason, overstating tax on both sides of the comparison and exaggerating the apparent benefit of acting early. Indexing removes that artefact.
The OAS clawback is modelled as a separate charge layered on top of regular income tax, not as part of the bracket table.
| Parameter | Value | Treatment |
|---|---|---|
| Minimum recovery threshold | $95,323 | 2026 figure; indexed to inflation each projection year |
| Recovery rate | 15% | Applied to net income above the threshold |
| Maximum recovery | 100% of OAS | Clawback is capped at the OAS actually received that year |
Because clawback is charged at 15 cents per dollar of income above the threshold and stacks on top of the marginal rate already applying at that income level, the effective marginal rate in the clawback zone is materially higher than the headline bracket rate. That interaction is the reason the withdrawal ceiling in the optimized plan is set relative to the clawback threshold rather than to a tax bracket.
When the model needs a registered withdrawal to produce a specific amount of after-tax cash, it solves for the gross amount by bisection over forty iterations rather than grossing up at a single marginal rate.
A one-pass gross-up is wrong whenever a withdrawal spans more than one bracket or crosses the clawback threshold — it systematically under-draws, sometimes by thousands of dollars, and the shortfall compounds through the rest of the projection. After-tax cash increases monotonically with the gross draw, so bisection converges to the exact figure.
You enter the annual amounts you would receive at 65. The model adjusts those figures for whatever start age it is testing, using the actual statutory adjustment factors.
| Benefit | Before 65 | After 65 | Range tested |
|---|---|---|---|
| CPP | −0.6% per month (−7.2% per year) |
+0.7% per month (+8.4% per year) |
60 to 70 |
| OAS | No early start available | +0.6% per month (+7.2% per year) |
65 to 70 |
Taken at the extremes, CPP started at 60 is 36% below the age-65 amount, and CPP deferred to 70 is 42% above it. OAS deferred to 70 is 36% above the age-65 amount. These are permanent adjustments, and both benefits are then indexed to the inflation rate you specify for the remainder of the plan.
OAS payments receive a permanent 10% increase beginning at age 75, introduced in 2022. The model applies this on top of any deferral adjustment, so someone who defers OAS to 70 and reaches 75 receives both the deferral premium and the age-75 increase.
The optimizer only considers start ages at or after your stated retirement age. If you retire at 66, CPP at 60 is not on the menu, and the search space narrows accordingly. This reflects the fact that the model does not simulate employment income, so it has no way to represent someone drawing CPP while still working.
Quebec residents: the model applies CPP rules and terminology throughout. QPP adjustment factors are broadly similar but not identical, and QPP now permits deferral past 70. Quebec results should be treated as an approximation on the benefit-timing dimension.
Two regulatory schedules constrain registered withdrawals: a floor that begins at 71, and a ceiling that applies to locked-in money throughout.
From age 71, the model applies the standard CRA minimum withdrawal percentages to the combined RRSP and LIRA balance, treating both as converted to a RRIF and LIF respectively. The schedule runs from 5.28% at 71 to 20% at 95 and above.
The minimum is a floor, not a target. In the optimized plan, the model takes exactly the minimum from 71 onward and funds any remaining spending from other accounts. Active melting happens before 71, which is the entire point of the strategy: the withdrawal is voluntary until it isn't.
LIRA balances are subject to an annual maximum withdrawal once converted to a LIF. The model caps the LIRA portion of any registered draw at the applicable LIF maximum percentage for that age, and pushes the excess demand to the RRSP, which has no withdrawal ceiling.
The table used covers ages 50 through 90, rising from roughly 6.3% at 50 to 100% at 89 and above. It reflects the 2026 schedule for the Ontario, Alberta, British Columbia, New Brunswick, Saskatchewan, and Newfoundland and Labrador grouping — the most common one.
Where this can overstate flexibility: LIF rules follow the regulator of the originating pension plan, not your province of residence. Federally regulated LIFs and those governed by Manitoba, Nova Scotia, and Quebec rules are more restrictive than the table used here. If your locked-in money falls under one of those regimes, the model gives your LIRA more room than you actually have, and the optimized plan will be modestly harder to execute than it appears.
Several provinces also permit the greater of the table percentage or the prior year's investment return. At typical return assumptions the table percentage dominates before age 80, so the model uses the table alone as a single approximation.
RRSP and LIRA withdrawals are taxed identically, so the split between them affects only the LIF constraint. The model allocates a target registered draw pro-rata across the two balances, applies the LIF cap to the LIRA share, and reassigns any capped-out amount to the RRSP where capacity allows.
For each year of the projection, the optimized plan resolves cash needs in a fixed order. The order carries the strategy: registered money is drawn first by design, not because it is cheapest in the moment.
There is a real tension here, and the model resolves it by testing both answers rather than assuming one.
The intuitive ceiling is the OAS recovery threshold: withdraw up to it, never past it, never trigger a clawback. That is the standard mode, and the model sets the target at $500 below the threshold in years when OAS is being received.
But stopping at the threshold can be the wrong answer. If the RRIF balance arriving at 71 is large enough, mandatory minimums will push income above the threshold every year from 71 to the end of the plan — producing a clawback stream far larger than the one avoided by being cautious in the years before. Accepting some clawback at 67 to eliminate a clawback cliff at 71 through 92 can be a clearly better trade.
So the model tests an aggressive mode as well, lifting the pre-71 ceiling above the clawback threshold:
| Condition | Ceiling | Reasoning |
|---|---|---|
| Standard year | 120% of threshold | Beyond roughly 20% above the threshold, the combined marginal rate plus clawback rarely beats the deferred tax being avoided |
| Year deferred CPP begins | 105% of threshold | Deferred CPP can be up to 42% larger than the age-65 amount; stacking a meltdown draw, new CPP, and new OAS in one year produces a spike worth avoiding |
| Fewer than 2 OAS years before 71 | Not applied | With a one-year window, aggressive mode dumps everything into a single year, which is rarely optimal |
Both modes are run for every CPP and OAS start-age combination the optimizer considers, and whichever produces more after-tax wealth is what you see. You are never asked to pick.
The model does not use a rule of thumb for benefit timing. It runs the full projection for every viable combination and picks the winner on the numbers.
CPP start ages from 60 to 70, OAS start ages from 65 to 70, each filtered to ages at or after retirement, crossed with both meltdown intensities. For someone retiring at 60, that is 11 CPP ages × 6 OAS ages × 2 modes = 132 complete lifetime projections, each running year by year to the end of the horizon and through the terminal tax calculation.
This is why the answer sometimes surprises. Benefit timing interacts with the meltdown window in ways that are hard to reason about from the outside: deferring OAS to 70 creates six extra years of clean melting room below the clawback threshold, but it also shortens the window in which the aggressive mode is worth using. The search finds these interactions; intuition often doesn't.
Each candidate plan is scored as:
What the objective implies: maximizing after-tax terminal wealth is an estate-oriented goal. If your actual priority is maximizing guaranteed lifetime income, or insuring against living to 100, or spending down deliberately to zero, a different objective would produce a different answer — most obviously on CPP timing, where deferral is partly longevity insurance rather than a tax play. The model optimizes for after-tax wealth at your stated end age. It has no view on how long you will live.
Non-registered withdrawals are the only funding source where the tax depends on what you paid for the assets, so the model tracks cost base explicitly rather than assuming a rate.
You enter both the market value and the adjusted cost base. The difference is the embedded gain. When a withdrawal is made, the model treats it as a proportional disposition: if 40% of the account's value is unrealized gain, then 40% of any withdrawal is a realized capital gain, and the ACB is reduced by the remaining 60%.
Realized gains are included in taxable income at 50%. This is the rate in effect for 2026. The two-thirds inclusion rate proposed in 2024 never came into force and was cancelled in March 2025.
Between withdrawals, the account grows at the retirement return assumption while the ACB stays fixed — so the embedded gain compounds, exactly as it does in reality. An account left untouched for twenty years arrives at the terminal calculation with a far larger taxable gain than it started with.
A gross-up is applied when drawing to meet a cash target, since part of each dollar withdrawn is consumed by tax on the gain it realizes. The gross-up factor is bounded so that no draw is inflated by more than a factor of two, preventing an unstable result in edge cases where the embedded gain approaches the entire account value.
After 71, mandatory RRIF minimums frequently deliver more cash than the plan needs to spend. What happens to the excess turns out to matter a great deal for the fairness of the comparison.
Each year, the model reconciles exactly: total cash received, less tax paid, less spending. If the result is positive, the surplus is deposited into the non-registered account with an adjusted cost base equal to the amount contributed — so it is not taxed a second time when later withdrawn.
Without this, surplus cash would simply disappear from the projection. That would penalize the unplanned path twice over: once through the tax on forced withdrawals it didn't need, and again by deleting the after-tax proceeds entirely. The unplanned path's forced withdrawals are inefficient, but the money still exists and still compounds. Modelling that honestly makes the comparison harder to win, which is the right standard for a number the calculator puts in front of you as a headline.
At the end of the planning horizon, the model applies the deemed disposition rules that apply on death for a person with no surviving spouse.
| Account | Treatment |
|---|---|
| RRSP and LIRA | Entire remaining balance added to taxable income in the final year, taxed at the resulting marginal rates |
| Non-registered | Capital gain on market value less ACB, included at 50% and added to final-year income |
| TFSA | Passes to the estate tax-free |
Tax is computed on the combined final-year income — fixed benefits plus the full registered balance plus the included capital gain — using brackets inflated to the final year. The tax already attributable to that year's ordinary income is then subtracted, so nothing is double-counted against the lifetime figure.
Using final-year inflated brackets matters. Applying today's nominal brackets to a balance that has compounded for thirty years would overstate terminal tax substantially, and would do so more for the unplanned path than the optimized one — inflating the headline savings figure for the wrong reason.
Two simplifications worth knowing about. First, no spousal rollover: in reality, registered assets transferring to a surviving spouse defer this tax entirely until the second death. A married couple's terminal tax bill arrives later and, depending on ages, can be materially different. Second, the terminal calculation applies income tax but does not add a further OAS recovery charge on the deemed-disposition income in the year of death.
Both simplifications sit on the same side of the ledger for a couple: the model shows the single-person outcome, which is the more severe one. Treat the terminal figure as the tax a single person would face, and as an upper bound on the immediate bill for someone with a surviving spouse.
If your current age is below your retirement age, the model first grows your balances forward to the retirement date before the withdrawal projection begins.
All inflation indexing — spending, CPP, OAS, the clawback threshold, and tax brackets — runs from your current age, not from your retirement age. If you are 55 and retiring at 65, spending has already been inflated by ten years when the withdrawal projection starts.
This is what makes the inputs mean what you expect. The $60,000 you enter as retirement spending is $60,000 of today's purchasing power, not $60,000 of nominal dollars a decade from now.
The accumulation phase models investment growth only. Employment income, tax on that income, RRSP deduction refunds, and TFSA contribution room limits are all outside the model. It will not stop you entering a TFSA contribution above the annual limit, and it does not credit you for the refund generated by an RRSP contribution.
The headline number is the difference in total tax — lifetime plus terminal — between the optimized plan and the unplanned baseline. It is worth being precise about what drives that gap.
The figure combines two distinct decisions, not one. The baseline takes CPP and OAS at 65 and makes no voluntary registered withdrawals. The optimized plan changes both: it selects benefit start ages by search and melts down the registered accounts. The savings figure is the joint effect.
For most inputs the sequencing decision does the heavier lifting, but not always — for someone with a large DB pension and a modest RRSP, benefit timing can dominate. The year-by-year table and the charts below the result let you see which lever is doing the work in your case: compare the withdrawal columns before 71 to see the meltdown, and check the selected CPP and OAS ages to see the timing effect.
What the figure is not is a forecast. It is the gap between two internally consistent scenarios under one set of assumptions about returns, inflation, longevity, and tax law — every one of which will be wrong to some degree. The order of magnitude of the gap is the useful output. The precise dollar figure is not.
Run the calculator →Everything the model leaves out, in one place. Several of these are consequential enough to change the answer for particular households, so it is worth reading the list against your own situation rather than skimming it.
What this model is for. It is built to show the order of magnitude and the direction of withdrawal sequencing decisions for Canadian households with substantial registered assets — to answer whether this is a $5,000 question or a $200,000 one, and which way the answer points. It is not built to produce a plan you can execute without further advice, and it will not match the output of professional planning software that models your credits, your spouse, and your actual holdings.
This calculator is provided for general informational and educational purposes only. It does not constitute financial, tax, legal, or investment advice, and no advisory or fiduciary relationship is created by using it. The output is intended to illustrate the order-of-magnitude impact of withdrawal sequencing decisions — not to produce a precise plan for any individual. Before acting on anything you see here, consult a qualified fee-only financial planner or tax professional who has reviewed your full personal circumstances, including risk tolerance, health, family situation, estate goals, and the specifics of every account you hold.
The author and publisher make no representations or warranties, express or implied, regarding the accuracy, completeness, or fitness for any particular purpose of the information produced by this calculator. To the fullest extent permitted by applicable law, the author and publisher disclaim all liability for any loss, damage, or negative consequence of any kind arising directly or indirectly from the use of or reliance on the calculator's output. Tax rules, benefit programs, and contribution limits referenced here may change after publication.