The starting rule: no ordinary tax between Canadian corporations
The reason intercorporate dividends are tax-free is that the profit behind them has already paid corporate tax once, and Parliament decided it should not pay it again just because it moved up a corporate chain. The mechanics are a deduction, not an exemption: the receiving corporation includes the dividend in income, then deducts the same amount under section 112, leaving nothing behind for regular corporate tax. The result is that after-tax profit can travel from an operating company to a holding company, or through several tiers of a corporate group, without shrinking at each step.
That one deduction is what makes the classic owner-managed structure work. The operating company earns profit and pays tax at its rate. The surplus moves up to the holding company as a dividend, free of tax, where it is invested or held safely away from the business's creditors. Personal tax only arrives when money finally leaves the corporate system and lands in a shareholder's hands. Deciding when that final step happens, and in what form, is the heart of corporate tax planning for owner-managed businesses.
The rule is also broader than cheques marked dividend. Tax law treats several corporate events as deemed dividends, redeeming shares, winding up a company, reducing paid-up capital beyond its tax value, and when those flows run between corporations they ride the same section 112 machinery and face the same three exceptions below. If your group is redeeming shares as part of a reorganization or buyout, the analysis on this page applies just as much as it does to an ordinary annual sweep of profit.
Everything else on this page is about the three exceptions, because each one exists to stop a specific abuse, and each one occasionally catches an honest structure that was not paying attention.
Exception one: Part IV tax on dividends from companies you are not connected to
Part IV tax exists to stop private corporations from being used as tax shelters for portfolio dividends. Without it, an owner could hold public-company shares inside a corporation, collect dividends that section 112 wipes clean, and defer personal tax indefinitely. So when a private corporation receives a dividend from a corporation it is not connected to, a bank stock, a public company, a private company where it holds only a small stake, it pays Part IV tax of 38⅓% of the dividend.
The saving grace is that Part IV tax is refundable, not permanent. It goes into the corporation's refundable dividend tax on hand, RDTOH, a notional account that tracks refundable taxes the corporation has paid on investment income. When the corporation later pays taxable dividends to its own shareholders, it recovers a dividend refund of 38⅓% of the dividends paid, until the pool runs dry. The design is a toll and a rebate: tax is prepaid while the money shelters inside the corporation, and handed back when the money continues its journey out to a person, who pays personal tax on it. Since 2019 the pool is split into an eligible and a non-eligible portion, which controls which kind of dividend releases which refund, a detail that matters when we plan payouts but not for understanding the system.
For an owner-managed group, exception one is mostly about the holding company's portfolio. Dividends from the brokerage account trigger Part IV tax every year, and those balances sit waiting until the owner takes taxable dividends personally. A payout plan that ignores a loaded RDTOH pool is leaving the corporation's own money parked with CRA.
Exception two: a connected payer that receives a dividend refund
Even between connected companies, a dividend carries Part IV tax with it when the payer gets a refund for paying it. The rule is mechanical: if the paying corporation receives a dividend refund out of its own RDTOH because of the dividend, the receiving corporation pays Part IV tax equal to its share of that refund. The refundable tax is not erased by moving the money up a tier; it climbs the chain along with the cash, and it keeps climbing until it is finally released by a dividend to an individual, who pays real personal tax.
Picture an operating company that realized a large capital gain, paid the high refundable rate on the taxable half, and now dividends the proceeds up to its holding company. The opco collects its refund, and the holdco inherits an equivalent Part IV liability, adding to its own RDTOH. Nothing has been lost, but nothing has escaped either. Owners are sometimes startled by a Part IV bill on what they understood to be a tax-free intercompany dividend; the answer is almost always that the payer had refundable tax balances, and the system is simply passing the toll along.
Timing adds a wrinkle worth knowing about. The payer's dividend refund is computed for the payer's own taxation year, so the Part IV consequence of a dividend received in January may not be fully knowable until the payer's year closes months later, and the two companies often have different year-ends. Groups that pay dividends casually through the year discover these interactions at filing time; groups that plan them declare each dividend with the payer's refund position already projected, which is a spreadsheet exercise, not a hardship.
Whether your companies are connected in the first place has a precise two-part test, control or the ten-percent votes-and-value threshold, and it is worth understanding on its own: see how connected corporation dividends work. For this page, the point is simpler: connection turns off Part IV tax except to the extent of the payer's refund.
Exception three: subsection 55(2), when a dividend is really a capital gain
Subsection 55(2) exists because a tax-free intercorporate dividend can be used to dodge capital gains tax, and CRA polices the boundary hard. The move it targets looks like this: your holding company is about to sell the operating company's shares for a large gain, so just before the sale the opco pays a huge dividend up to the holdco. The company is now worth less, the sale price drops, and the gain shrinks, value that would have been taxed as a capital gain slipped out as a tax-free dividend instead. Where one of the purposes of a dividend is a significant reduction of a capital gain like that, subsection 55(2) recharacterizes the dividend as a capital gain, with tax to match.
The main shelter from the rule is safe income: retained, already-taxed earnings that have accumulated while you held the shares. A dividend paid out of safe income on hand is protected, because it is exactly the after-tax profit the section 112 deduction was designed for, moving it up does not dodge anything. The catch is that safe income is a computed number, built from the corporation's tax history year by year with many adjustments, and it is rarely equal to retained earnings on the balance sheet. Paying a large dividend on the assumption that safe income covers it, without the computation, is how honest reorganizations acquire six-figure problems.
The practical rules of thumb: routine annual sweeps of profit from opco to holdco are rarely troubled, because they track safe income as it accrues. Large, unusual dividends deserve a safe income computation first, especially with a sale, a reorganization or a new shareholder anywhere on the horizon. And Part IV tax and subsection 55(2) interlock, a dividend that bears Part IV tax without a refund back to the payer's group is generally outside 55(2)'s reach, which is one of several reasons the analysis belongs in professional hands before the resolution is signed, not after. For deemed dividends on share redemptions the test is stricter still, looking at results rather than purposes, so redemptions inside a group get planned with particular care.
What arrives at the other end: the dividend keeps its character
A dividend does not just carry cash between your companies, it carries its tax character, and the receiving corporation inherits it. Eligible dividends, the kind sourced from income taxed at the general corporate rate, land in the recipient's GRIP pool and can be paid out again as eligible dividends, which carry a lower personal rate when they finally reach you. Non-eligible dividends, sourced from income that enjoyed the small business deduction, stay non-eligible. Capital dividends, the tax-free kind, keep their tax-free character as they pass between private corporations, so an opco's capital dividend account can travel up to the holdco and out to you intact. The character rules matter because they set the personal tax bill years later, at the final step out of the corporate system.
| Dividend received by your company | Corporate tax on arrival | What it does to the recipient's accounts |
|---|---|---|
| Taxable dividend from a connected payer, no refund to payer | None; section 112 deduction applies | Adds to GRIP if eligible; otherwise available for non-eligible payout |
| Taxable dividend from a connected payer that gets a dividend refund | Part IV tax equal to the recipient's share of the refund | Adds to the recipient's RDTOH, refundable on its own future payouts |
| Portfolio dividend from a non-connected corporation | Part IV tax at 38⅓% | Adds to RDTOH; recovered when taxable dividends are paid out |
| Dividend caught by subsection 55(2) | Recharacterized as a capital gain, taxable half taxed as investment income | Untaxed half credits the capital dividend account; refundable tax on the taxed half |
| Capital dividend from a private payer | None, and none later | Credits the recipient's capital dividend account for tax-free payout |
Notice what the table implies about the income behind the dividends. Profit that was active business income in the operating company arrives at the holding company already sorted into eligible or non-eligible character, and that sorting, not anything the holding company does, decides the personal rate on the eventual payout. Getting money between your companies is the easy half; getting the character right is where the planning value lives.
The facts that change the answer, and how we handle it
Five facts decide how any particular intercompany dividend is taxed, and we check all five before it is declared. Whether the companies are connected, under the control test or the votes-and-value test, at the time the dividend is received. Whether the payer has RDTOH balances, because a refund to the payer means Part IV tax to the recipient. How large the dividend is against the payer's computed safe income, the subsection 55(2) question. What character the dividend carries, eligible, non-eligible or capital, and what that does to the recipient's pools. And what the dividend is for, a routine sweep, a purification, a creditor-proofing move or a pre-sale step, because purpose is precisely what the anti-avoidance rules read.
None of this argues against moving money between your companies; it argues for declaring dividends deliberately, with the accounts computed first and the paper trail complete, including the T2 schedule that reports dividends between connected corporations. For a corporate tax planning CPA in Ontario this is standing work: we track the RDTOH, GRIP and capital dividend pools through the year inside our corporate tax engagements, and compute safe income before any dividend large enough to matter. If a reorganization or a sale is in motion, the dividend plan becomes part of that project. Either way, the review costs a fraction of what an unplanned recharacterization does, and it starts with a free 15-minute discovery call.
