Thesis of the Century (STF Theme 69).
Exclusion of ICMS from the PIS/COFINS base — RE 574.706 and the 2021 modulation.

The trajectory of the ruling

Origin of the thesis

The discussion dates back to the 1990s. The central controversy: is the ICMS, although charged separately on the invoice, recorded accounting-wise as gross revenue by the company? If so, it is part of the PIS/COFINS base (the cumulative regime of Law 9,718/98 or the non-cumulative regime of Laws 10,637/2002 and 10,833/2003); if not, it must be excluded as it is not the company's own revenue.

The stakes were structural rather than marginal. PIS and Cofins are calculated on revenue or turnover, and for decades the Federal Revenue treated the full invoice amount — ICMS included — as the taxable base. Because every sale in Brazil carries ICMS that is highlighted on the invoice and later passed on to the State, including that amount inflated the base of two federal contributions across the entire economy, year after year. The thesis therefore reached far beyond any single sector: every company that sells goods subject to ICMS and pays PIS/Cofins on revenue was, in principle, overpaying. That breadth is exactly why the dispute became known as the "Thesis of the Century" — not for the novelty of the legal argument, but for the sheer size of the universe of taxpayers and the volume of money at issue.

Procedural trajectory

• RE 240.785 (2014): the first favorable STF precedent for the exclusion, but without binding effect — an individual case
• Recognition of general repercussion: September 2008 (Theme 69)
• Merits ruling: March 15, 2017 — the STF, by 6 votes to 4, sets the exclusion
• The Union's modulation request: filed in May 2017
• Modulation decision: May 13, 2021 — effects from 03.15.2017 (the date of the ruling), except for suits and administrative proceedings filed up to that date, which keep a retroactive right of 5 years from filing

Composition of the merits ruling (2017)

The vote was 6 to 4, with the following composition:

• Pro-taxpayer (6): Justices Cármen Lúcia (rapporteur), Rosa Weber, Marco Aurélio, Ricardo Lewandowski, Celso de Mello and Luiz Fux
• Pro-Treasury (4): Justices Edson Fachin, Roberto Barroso, Dias Toffoli and Gilmar Mendes — the dissenters held that the relevant figure was the ICMS collected, not the ICMS highlighted on the invoice

Thesis set

"ICMS does not form part of the calculation base for the levy of PIS and COFINS." (Justice Cármen Lúcia, rapporteur)

Central legal basis: the constitutional concept of "revenue" or "turnover" (Federal Constitution, art. 195, I, b) does not cover amounts that merely pass through the company's accounting on their way to another entity — ICMS is the State's revenue, not the taxpaying company's. Application of the ability-to-pay principle (Federal Constitution, art. 145, §1) — what does not belong to a person is not taxed against that person.

The practical reach of the wording chosen by the rapporteur only became clear later, in the clarification motions. Two questions had to be answered before the thesis could be operated: which ICMS figure should be excluded — the amount highlighted on each invoice or the smaller amount effectively collected after the State's own non-cumulativity credits — and from what date the decision would produce effects. The first question was settled in favor of the ICMS highlighted on the invoice, the larger of the two figures; the second was the subject of the 2021 modulation. Both answers are decisive for any recovery calculation, because they define, respectively, the size of the credit and the period over which it can be claimed.

Modulation (05.13.2021) — who still has a retroactive right

Why there was a modulation

The Union argued that a retroactive application of Theme 69 would create a catastrophic fiscal impact — an estimate of BRL 250-350 billion in refunds of PIS/COFINS collected from 1996 to 2017. The STF, weighing legal certainty and budgetary impact, modulated the effects of the decision.

The modulation rule

The 05.13.2021 decision established:

• Effects of the thesis: from 03.15.2017 (the date of the merits ruling), the exclusion is mandatory for all taxpayers
• Retroactive exception: taxpayers that already had a lawsuit filed by 03.15.2017 keep the right to recover PIS/COFINS retroactively, up to 5 years before filing (National Tax Code, art. 168)
• Those who filed after 03.15.2017: have only a prospective right (from the date of the ruling — they can recover from 03.15.2017 onward)

The "suit filed by 03.15.2017" criterion

The cut-off criterion is the filing of the suit, not a favorable decision. Any lawsuit — writ of mandamus, ordinary action, tax-foreclosure objection — filed by that date and discussing the matter of Theme 69 counts, even if still pending. What matters is the protocol date: a taxpayer who had already taken the matter to court before the merits ruling is treated differently from one who only acted afterwards, regardless of how each case ultimately fared. This is why the diagnosis of any company always begins with a litigation check — confirming whether an action existed before 03.15.2017 is the single fact that separates a full retroactive recovery from a prospective-only one, and it cannot be assumed; it has to be verified against the court records.

What was settled

• The ICMS highlighted on the invoice (and not the ICMS collected/paid) is the amount to be excluded — set in RE 574.706 and confirmed in clarification motions
• Application to all taxpayers — regardless of size or tax regime (Actual Profit, Presumed Profit or Simples Nacional within the sub-limit)
• Usage modality: administrative offset via PER/DCOMP (with prior qualification if obtained judicially) or direct imputation in the current assessment

How to calculate and use the credit

What to exclude

The taxpayer must exclude from the PIS and COFINS base the value of the ICMS highlighted on each exit invoice. It is not the ICMS paid at the end of the period (after credits), but the value highlighted on each individual operation.

The Federal Revenue has itself incorporated the thesis: IN RFB 2,121/2022 (art. 26, XII) provides that the ICMS highlighted on the tax document does not form part of the PIS and Cofins base, aligning with the STF modulation. Note one limit in the sole paragraph: the exclusion does not reach the ICMS highlighted on sales with suspension, exemption, zero rate or non-incidence. In other words, the prospective application has express support in the administration's own rule, which reduces the risk of disallowance.

Simplified calculation

Example: a retail company with monthly revenue of BRL 10 million, average state ICMS rate of 18%:

• Gross revenue: BRL 10,000,000
• ICMS highlighted on invoices: BRL 1,800,000 (18%)
• PIS/COFINS base before the thesis: BRL 10,000,000
• PIS/COFINS base after the thesis: BRL 8,200,000
• Monthly saving (non-cumulative regime, combined rate 9.25%): BRL 166,500
• Estimated annual saving: BRL 1,998,000

Usage modality

There are two paths:

1. Direct prospective application — assess PIS/COFINS already excluding ICMS. It does not require judicial authorization — the consolidated thesis applies
2. Retroactive recovery via PER/DCOMP — requires:
   • A writ of mandamus with a final and unappealable decision recognizing the right (STJ Summaries 213 and 460)
   • Prior qualification of the credit at the Federal Revenue (administrative procedure)
   • After qualification, monthly filing of a DCOMP to offset against taxes coming due

Two notes keep the calculation defensible. First, the exclusion is taken operation by operation, on the ICMS highlighted in each exit document, and not on the net ICMS settled at the end of the month — applying the net figure understates the credit and departs from the precedent. Second, the rule of IN RFB 2,121/2022 cuts both ways: it confirms the exclusion for ordinary taxed sales, but its sole paragraph removes from the calculation any ICMS highlighted on operations under suspension, exemption, zero rate or non-incidence. A clean credit therefore depends on classifying each operation correctly, which is feasible only when the underlying tax data is read in full rather than estimated from aggregate balances.

Digital tax audit

A 5-year retroactive calculation requires invoice-by-invoice analysis of hundreds of thousands of operations. The TaxUp team operates with a digital tax-audit pipeline that processes SPED Fiscal and electronic invoices to reconstitute the historical calculation — doing it manually would be unfeasible. The same pipeline produces the evidence the Federal Revenue requires at the prior-qualification stage: a month-by-month demonstration of the base that was used, the ICMS that should have been excluded, and the resulting overpayment updated by the Selic rate. A credit that is well documented at this stage moves through qualification and offsetting; one built on rounded estimates is the most common reason for rejection.

"Offspring" theses — derived from the Thesis of the Century

The consolidation of Theme 69 opened the way for multiple derived theses applying the same reasoning (excluding taxes transferred to third parties from the base of other taxes):

Offspring thesis 1 — Exclusion of ISS from the PIS/COFINS base

A literal application of the Theme 69 reasoning to ISS. It is on trial at the STF — Theme 118 (RE 592.616), tied 5×5, awaiting the tie-breaking vote of Justice Luiz Fux; it was scheduled for February 2026 and then removed from the docket with no new date. Service-provider companies (law, consulting, IT) that paid PIS/COFINS on gross revenue including ISS have a valid thesis. Caution: the STF has not yet decided definitively, and a pro-Treasury outcome is a real risk — a prospective writ of mandamus is advisable to suspend enforceability while awaiting the ruling.

Offspring thesis 2 — Exclusion of PIS/COFINS from its own base

A self-exclusion thesis: PIS and COFINS should not be levied on themselves (the "inside" calculation). The thesis is pending at the STF (Theme 1.067, RE 1.233.096), with no trial date; there are favorable decisions in lower courts, but no binding thesis yet — a medium-risk bet.

Offspring thesis 3 — Exclusion of ICMS-ST from the PIS/COFINS base

Application to the Tax Substitution regime. STJ Theme 1.125 is already decided in favor of the substituted taxpayer (who can exclude the ICMS-ST from the PIS/COFINS base), with its own modulation of effects. For taxpayers subject to a significant ST regime, it is an already-mature recovery opportunity.

Offspring thesis 4 — Exclusion of IPI from the PIS/COFINS base

The same reasoning applied to IPI, which is already highlighted separately from the industrial company's revenue — a discussion generally handled administratively, without a consolidated binding repetitive precedent; a more executive than judicial profile. Its viability depends on a case-by-case analysis.

Attention — an offspring thesis that FAILED: ICMS in the IRPJ/CSLL base (Presumed Profit)

Not every offshoot prevailed, and that matters as much as knowing the ones that did. The attempt to apply the Theme 69 reasoning in order to exclude ICMS from the IRPJ and CSLL base under the presumed profit was rejected by the STJ in Theme 1008 (REsp 1,767,631/SC and 1,772,470/RS, rapporteur Justice Regina Helena Costa, 2023, final and unappealable on 12/14/2023): it was held that "ICMS forms part of the calculation base of the IRPJ and the CSLL when assessed under the presumed-profit system" — because the presumed base is a percentage of gross revenue, with no right to deduct costs or taxes. It is the reminder that the "parent" won, but not every offspring follows the same fate: each thesis demands its own analysis of admissibility.

Each offspring thesis requires a specific analysis of admissibility, procedural viability and the risk-benefit balance. The TaxUp team works with case-by-case modeling.

Who can still benefit (2026)

More than seven years after the merits ruling and over four years after the modulation, the thesis is no longer a question of whether ICMS is excluded — that is settled and even reflected in the Federal Revenue's own rule. What remains is a question of who still has an open window and how much of the credit is still within reach. Four situations cover almost every company, and the right strategy depends on which one applies.

Scenario 1 — Company that filed by 03.15.2017

It keeps the right to retroactive recovery:

• 5 years before filing (the general rule of National Tax Code, art. 168)
• It should already be in the phase of judgment enforcement or qualification at the Federal Revenue
• A typically significant credit — large companies may have BRL 100 million+ to recover

Scenario 2 — Company that did NOT file by 03.15.2017

A prospective right only:

• It can apply the exclusion in the current assessment (no need for a lawsuit)
• It can seek recovery from 03.15.2017 — 5 retroactive years counted from the final decision of any prospective writ of mandamus
• A consolidated thesis — zero risk of assessment for direct application

Scenario 3 — A new company or one that never used it

It can still recover PIS/COFINS overpaid in the last 5 years:

• By the consolidated thesis (Theme 69), retroactive application is possible since 03.15.2017 (the National Tax Code limitation period)
• A digital tax audit reconstitutes the amounts
• Calculation, a writ of mandamus for qualification, monthly offset DCOMP

Scenario 4 — Offspring theses

Companies that already used Theme 69 can now structure offspring theses — especially ISS (STF Theme 118, still on trial — tied 5×5, no thesis set). The ICMS-ST (STJ Theme 1.125) is already consolidated and modulated to 03/15/2017. For ISS, a prospective writ of mandamus suspends enforceability while there is no definitive decision.

Current risks

• Limitation period: 5 years counted from the undue payment. A 2018 credit is time-barred in 2023 — a company that does not act loses it definitively
• Rigorous prior qualification: the Federal Revenue has intensified its analysis. A poorly done calculation = rejection + lost time
• Attention to the modulation: applying undue retroactivity (filing after 03.15.2017 and claiming 5 years before) is a common cause of disallowance

The window is closing: PIS and Cofins end in 2027

There is a clock running that changes the math for anyone who has not yet acted. Under LC 214/2025, PIS and Cofins are extinguished in 2027, replaced by the CBS (in 2026 there is only the 0.9% CBS test rate; in 2027 it becomes full and the old contributions are withdrawn). For the Thesis of the Century, that has two direct consequences:

First, the recoverable stock stops growing: with no new PIS/Cofins collections after 2027, the credit no longer renews itself, while the five-year limitation keeps knocking down the oldest months. Second, the main "destination account" disappears: the recognized credit remains offsettable against other federal taxes (IRPJ, CSLL, social-security contributions), but the natural destination — the PIS/Cofins liabilities themselves — vanishes. Surveying and filing now maximizes what is still recoverable and leaves time to use the credit before the turn. Whoever waits loses on both sides.

Practical case: an industry recovering Theme 69

An illustrative scenario (figures fictional and rounded). An actual-profit industrial company (non-cumulative PIS/Cofins, 9.25%), with gross revenue of BRL 120 million per year, highlights on its exit invoices about 18% of ICMS — BRL 21.6 million per year that were unduly included in the PIS and Cofins base.

The credit is direct: 9.25% over BRL 21.6 million = about BRL 2.0 million per year. Over the five-year limitation period, that is approximately BRL 10 million of principal, which rises to BRL 12 to 13 million after the Selic adjustment. Because this industrial company did not file by 03/15/2017, it recovers the current five-year period — which shrinks with each month of inaction. The recommended route combines a writ of mandamus (which declares the right and secures the offset of the five years) with administrative use via PER/DCOMP against federal liabilities — sized already taking into account that PIS/Cofins will be extinguished in 2027.

The same company, were it on the presumed profit (cumulative PIS/Cofins at 3.65%), would recover proportionally less — about BRL 788k per year, ~BRL 3.9 million over five years plus Selic — but the right exists under both regimes. The figures are illustrative: the real amount depends on the invoice-by-invoice survey, by sector, product mix and State, which the TaxUp team performs through a digital tax audit over SPED and EFD-Contribuições.

How the TaxUp team acts on Theme 69

The TaxUp team acts on three fronts in Theme 69. Which front leads depends entirely on the diagnosis: a company with a pre-2017 action is already on the path to a large retroactive recovery and needs execution and qualification support; a company without one needs the prospective application secured and the current five-year window protected before more of it is lost to limitation. In both cases the work is the same in kind — reconstruct the base correctly, document it to the standard the Federal Revenue demands, and convert the recognized credit into actual offsetting against federal liabilities — but the sequence and the risk profile differ.

The three fronts:

1. Direct prospective application (no litigation)

• Diagnosis of applicability according to the client's tax regime (Actual, Presumed, Simples)
• Recalculation of the PIS/COFINS assessment of the last 5 years excluding the ICMS highlighted
• Amendment of ECF, EFD-Contribuições and DCOMP
• Application in the current assessment

2. Writ of mandamus and retroactive qualification

• Admissibility modeling (checking for a prior action or whether it is a new writ of mandamus)
• A writ of mandamus declaring the right to offset (STJ Summaries 213 and 460)
• After the decision becomes final, prior qualification before the Federal Revenue with a Selic statement
• Monthly PER/DCOMP operation until the credit is exhausted

3. Active offspring theses

• Specific admissibility analysis (Theme 118 ISS, Theme 1.125 ICMS-ST)
• A prospective writ of mandamus to suspend enforceability while the STF/STJ has not decided
• Monitoring the decision and adjusting the strategy

The fee model combines a fixed part (analysis + filing) and a variable part tied to the recovered credit (% of the qualified and offset amount). In high-value projects (estimated credit > BRL 5 million), a mostly success-fee model. A digital tax audit processes the historical SPEDs with no upfront cost in success-linked cases.