PHSP Qualification Tests Under the Income Tax Act

A private health services plan only gets its tax-free status because it clears a specific set of tests buried in subsection 248(1) of the Income Tax Act and fleshed out in CRA's Interpretation Bulletin IT-339R2. Miss one of those tests, and the reimbursements employees thought were tax-free turn into taxable income under paragraph 6(1)(a). This matters whether you call the arrangement a Health Spending Account, a cost-plus plan, or a Health Care Spending Account: the label changes nothing, and CRA applies the same tests regardless of what's printed on the plan document.

The statutory definition itself runs two tracks. Branch (a) covers a contract of insurance for hospital expenses, medical expenses, or both. Branch (b) covers a medical care insurance plan or hospital care insurance plan, or a combination of the two. Both branches carve out provincial and federal health insurance (OHIP, AHCIP, RAMQ, and the like), so those government plans can't double as a PHSP. That said, a plan built on top of or alongside provincial coverage isn't automatically disqualified just because it originated from a provincial framework. CRA's position is that if the plan meets the PHSP tests on its own merits, provincial origin doesn't taint it.

The five-element insurance test from IT-339R2

Paragraph 3 of IT-339R2 sets the bar: a PHSP has to be "a plan in the nature of insurance." Not a reimbursement scheme with an insurance-sounding name attached. Actual insurance, with all the structural pieces that come with it.

CRA breaks this into five elements, and every single one has to show up:

• An undertaking by one person. In a fully insured plan, that's the insurer. In a self-insured or cost-plus plan, the employer itself steps into that role and gives the undertaking.
• To indemnify another person. That's the employee, and any covered dependants, for a loss they actually suffer. This is indemnification, not a pre-loaded credit balance waiting to be spent.
• For an agreed consideration. In an employment context, that's the value of the employee's services. In an individual policy, it's the premium.
• For a loss or liability in respect of an event. Here, that means hospital or medical expenses covered under the plan.
• The happening of which is uncertain. This is where genuine insurance risk lives. If the event doesn't occur, no benefit gets paid. If it does, the plan pays.

That last element causes the most trouble in practice. A plan that guarantees reimbursement no matter what expenses come in, structured so payment is a near-certainty rather than a contingency, fails to clear the bar. If an employer simply credits a fixed dollar figure to an employee and pays out whatever gets submitted with zero risk on either side, that's not insurance. It's a salary top-up wearing an insurance costume.

Why the plan must carry a real element of risk, and what that rules out

CRA won't recognize a plan as a PHSP unless there's a real element of risk sitting somewhere in the structure, whether the risk-bearer is a third-party insurer or the employer itself acting as insurer under a self-funded arrangement.

For a cost-plus plan to hold up, several pieces need to be in place at once. The employer contracts with a trusteed plan or an insurance company to administer things. The employer commits to reimbursing the cost of claims plus an admin fee. The employment contract itself requires the employer to reimburse properly filed claims. And the arrangement includes a formal administrative relationship between the plan and the employees covered under it. Put together, the employer's total payout for any given period could land anywhere between zero, if nobody incurs eligible expenses, and the dollar ceiling set for that employee. That range, that genuine unpredictability, is the risk element the statute demands.

What kills it fast is a termination clause that lets the employer cancel the plan at any time, at its sole discretion, without notice. CRA treats this as a red flag, and for good reason: if the employer can walk away the moment a big claim occurs, there was never really an obligation to indemnify anyone. It was an option, not insurance.

There's an anti-avoidance dimension here too. CRA scrutinizes arrangements with no real risk baked in, particularly sole proprietors with no employees running a pay-as-you-claim structure that amounts to nothing more than reimbursing themselves for expenses they already knew they'd have. Cost-plus plans aren't inherently suspect. Structured correctly, they satisfy the insurance element just fine. The trouble starts when the plan's own terms erase the uncertainty the law requires.

The "all or substantially all" test: what qualifies as eligible coverage

Coverage composition matters just as much as structure. CRA's rule: a plan qualifies as a PHSP as long as all or substantially all, generally read as 90% or more, of the premiums or benefits paid relate to expenses eligible for the Medical Expense Tax Credit under subsection 118.2(2).

That threshold wasn't always 90%. Before November 2015, CRA held the line at 100% eligibility, so any non-eligible benefit, however minor, could jeopardize the whole plan. The shift to 90% was announced in November 2015 and applied starting with the 2015 tax year onward. In effect, this gave plans some breathing room: a small slice of non-eligible benefits, under 10% of the total, doesn't sink the plan anymore.

How the test gets measured depends on the plan type. For insured plans, CRA looks at premiums paid, not benefits actually paid out to employees in the year. For self-insured plans, the test runs the other direction: it looks at actual benefits paid to all employees over the calendar year. CRA's own published guidance offers an illustration where METC-eligible expenses represented 93% of total benefits paid in a year, which passed comfortably.

The core eligible categories line up with what qualifies for the METC: prescriptions, medical treatment, dental work, vision care, hospital expenses. Where plans run into trouble is scope creep. Gym memberships, personal development courses, and general wellness perks don't belong in a PHSP. They belong in a separate Wellness Spending Account, which carries entirely different tax treatment. CRA's position is that folding these items into a PHSP puts the plan's entire status at risk, not just the ineligible portion.

The formal plan requirement: documentation, limits, and binding obligation

A PHSP has to exist on paper. An informal understanding between employer and employee, however well-intentioned, doesn't meet the bar. There has to be a written document laying out what coverage is available, and the employer has to communicate the specifics to employees, including how much reimbursement they can expect and under what conditions.

Limits matter too. A PHSP typically has a ceiling on how much can be reimbursed, and a plan with no cap at all runs into trouble on two fronts. First, it starts to look less like insurance and more like open-ended compensation. Second, the Income Tax Act requires that any deductible outlay be reasonable given the circumstances, and an unlimited plan makes that reasonableness test hard to satisfy. Different employee classes can carry different annual limits within the same plan, which is both permitted and common; a senior executive tier and a junior staff tier don't need identical ceilings.

The termination clause issue comes back here too. A plan the employer can cancel at will, without notice, undermines both the contractual obligation and the insurance risk element at the same time, which is why CRA treats it as a compliance red flag rather than a minor drafting quirk.

Best practice for a corporation setting one of these up: write the reimbursement obligation directly into employment agreements, and back it with a board resolution documenting the plan's terms, limits, and eligible employee classes. And the plan needs to exist before claims start rolling in. Setting one up retroactively, after an employee already knows they've got a substantial dental bill coming, doesn't create a qualifying plan. It creates a paper trail CRA will see right through.

How incorporated businesses and shareholder-employees must satisfy the employee-capacity test

There's no statutory dollar cap on PHSP deductions for corporations. A corporation made up entirely of shareholder-employees can still set one up. But there's a catch that trips up a lot of small incorporated businesses: a shareholder's medical expenses only qualify for PHSP treatment if they're receiving the benefit as an employee, not as a shareholder collecting a perk tied to ownership.

CRA starts from a presumption that shareholders who can meaningfully influence company policy receive benefits because they own the business, not because they work for it. That presumption has to be rebutted, and CRA guidance, alongside IT-339R2, lays out three conditions to do it. The shareholder needs to be actively engaged in the business's day-to-day activities. The benefits provided need to be reasonable in scope and value. And the benefits need to look like something an arm's-length employee, doing similar work at a similarly sized company, would actually receive.

A written employment contract helps establish that employment relationship formally, strengthening the case that the benefit is received as an employee rather than as a shareholder. How the shareholder gets paid matters too. Someone compensated purely through dividends, with no T4 salary at all, has a weaker case for employee-capacity treatment than someone drawing at least part of their income as salary.

Get this wrong, and the consequence is blunt: the payment becomes a taxable benefit to the shareholder, and it's not deductible by the corporation either. That's double taxation on the same dollar, once as a lost deduction and once as personal income.

The case that illustrates this best is a Tax Court of Canada decision involving a company and its shareholder-employee (2004 UDTC 123, informal procedure). and Steve Cousins v. The Queen* (2004 UDTC 123, Tax Court of Canada, informal procedure). A cost-plus PHSP covered a shareholder-employee holding 51% of company shares, and the plan reimbursed a $35,996 knee surgery performed in a foreign country. The Tax Court ruled this was a shareholder benefit, not an employment benefit, largely because the sheer generosity of the payout made it implausible that an arm's-length employee doing similar work would ever have received the same coverage.

The sole shareholder / sole employee problem and the arm's-length benchmark

The single toughest scenario in this whole framework is the sole shareholder who's also the company's only employee. CRA has addressed this directly, more than once, and the answer isn't encouraging.

CRA has found that a cost-plus plan covering a sole employee-shareholder, their spouse, and other household members may not qualify as a PHSP, because it lacks the necessary elements of insurance discussed earlier. There's no one else in the company to compare against, so the risk element and the benchmark test both collapse.

CRA has reinforced this position in published guidance, noting that a self-insured HSA set up for a sole employee-shareholder and family members would likely fail to be a plan in the nature of insurance, and therefore wouldn't qualify as a PHSP at all.

The comparative test that has to be met asks whether non-shareholder employees, doing similar work at a similarly sized company, receive similar benefits under a similar plan. A sole shareholder-employee has no internal population to draw that comparison from, and this absence causes the issue, as shown by the comparative test requiring non-shareholder employees at a similarly sized company to be available for comparison.

One carve-out applies. CRA has confirmed that a plan covering a single plan member, where that member is the company's sole employee and deals with the company at arm's length, can still qualify. Arm's-length status is the deciding factor there, not headcount.

What this means practically: a sole owner-operator running their own incorporated business faces a real structural obstacle, and it's not a matter of one disallowed claim here or there. If CRA recharacterizes the plan, every single reimbursement made under it becomes a taxable shareholder benefit, retroactively. Documentation, an employment contract spelling out the reimbursement obligation, a board resolution, a notice requirement before any plan change, helps build the case that a genuine employment relationship exists. None of it eliminates the risk entirely. It just makes the argument stronger if CRA ever comes asking.

How unincorporated businesses qualify under section 20.01

Unincorporated businesses play by a different rulebook. Section 20.01 governs PHSP deductibility for sole proprietors and partners, and it's considerably more restrictive than the corporate route.

Two income tests apply under the rules for unincorporated businesses, and both need to be satisfied. First, the individual has to be actively engaged in the business on a regular, continuous basis, no passive ownership allowed. Second, one of two income conditions has to be met: either more than half of the individual's total income for the year comes from the business, or income from sources outside the business doesn't exceed $10,000 for the year.

Then there are hard dollar caps that corporations simply don't face. The deduction tops out at $1,500 for the proprietor, their spouse or common-law partner, and each household member over the age of 18. Household members under 18 are capped at $750 each.

None of this replaces the insurance test from IT-339R2. It stacks on top of it. The plan still has to be a genuine plan in the nature of insurance; section 20.01 just adds income eligibility rules and dollar ceilings on top of that baseline requirement.

Sole proprietors with no arm's-length employees run into the same wall as sole shareholder-employees in a corporation: no arm's-length relationship means no genuine risk element, and CRA's GAAR position treats a pay-as-you-claim structure in that setting as failing the insurance test outright. A sole proprietor who has at least one arm's-length employee is in materially better shape, since that relationship restores the benchmark comparison and the risk element the plan needs.

The contrast with incorporation is stark. No statutory dollar caps, no income tests, nothing close to the $1,500 ceiling facing unincorporated proprietors. For a professional carrying significant health expenses, that gap alone can make incorporation a decision worth running the numbers on.

Pulling the tests together: what a qualifying plan looks like in practice

A plan that actually holds up as a PHSP has to satisfy every single test at once, not just the ones that are easiest to document.

It needs the five insurance elements from IT-339R2: an undertaking, indemnification, agreed consideration, a covered loss, and genuine uncertainty about whether that loss occurs. It needs real risk baked into the structure, which means no termination-at-will clause that would let the employer walk away from obligations, and no structure that guarantees payout regardless of whether eligible claims arise. It needs at least 90% of its premiums or benefits tied to METC-eligible expenses, with wellness perks and gym memberships kept in a separate account entirely. It needs to exist in writing, with defined dollar limits, communicated to employees before any claims come in.

For incorporated businesses with shareholder-employees, it needs an employment contract, a salary component (not dividends alone), and benefits that would look reasonable to an arm's-length employee doing comparable work. Sole shareholder-employees face a benchmark problem that documentation can soften but not erase. For unincorporated businesses, it needs to clear the section 20.01 income tests and stay within the annual dollar caps, on top of the general insurance requirements everyone else has to meet.

Every one of these tests exists for the same underlying reason: to separate a genuine risk-sharing arrangement from a tax-advantaged way of paying salary. Structure the plan around real risk, real documentation, and a defensible employee relationship, and the tax-free treatment holds. Skip any one piece, and CRA has more than enough grounds, and more than one precedent, to treat the whole arrangement as taxable income instead.