Gratuity Tax Exemption: The Half-Month Formula Worked Through

Which table entry applies to you

The Act's salary deductions table in section 19(1)_Bill.pdf) has two gratuity rows.

• Sl. 5: gratuity received under the Payment of Gratuity Act, 1972. It is exempt to the extent it does not exceed the amount computed under sections 4(2) and 4(3) of that Act. We do not restate that Act's formula or eligibility rules here, because we have not verified them. Ask HR for the computation sheet.
• Sl. 6: any other gratuity. This is the three-way minimum worked below.

Both rows sit in the 2025 Act, which applies from 1 April 2026. A gratuity paid between then and 31 March 2027 falls in tax year 2026-27.

The three amounts in the Sl. 6 test

The exemption is the smallest of the three. The notified limit is not restated here, because we have not verified the current figure. Limb (c) is the only one that your exit date and pay history change, so the rest of this article concentrates on it.

Working limb (c)

Suppose you leave on 31 January 2027 after 7 years and 8 months of service. Your salary for this purpose was ₹60,000 a month until a revision to ₹66,000 on 1 October 2026. The ten months before the exit are April 2026 to January 2027.

The average is ₹62,400. Half of it is ₹31,200, and 7 completed years gives ₹2,18,400. The eight extra months do not count, because the table uses completed years.

Check with HR which pay elements your employer counts as "salary" for this limb. The worked figure treats ₹60,000 and ₹66,000 as that salary, not as gross pay.

Why the exit date moves the exemption

Two small date changes shift limb (c):

• Four more months. Leaving on 31 May 2027 instead completes the eighth year. It also moves more of the higher salary into the average. The ten months are August 2026 to May 2027: 2 at ₹60,000 and 8 at ₹66,000, averaging ₹64,800. Limb (c) becomes ₹32,400 × 8 = ₹2,59,200. That is above the ₹2,50,000 actually received, if the payout were unchanged, so the whole payout would be exempt.
• A revision just before exit. A raise in the last ten months lifts the average only in proportion to the months it covers.

That May 2027 date falls in tax year 2027-28, so the taxable part would be taxed at that year's rates, which are not yet known.

# Synthetic salaries and dates. Section 19(1) Table Sl. 6 logic only.
def gratuity_exempt(received, monthly_salaries_last10, completed_years,
                    notified_limit, exempt_used_before=0):
    assert len(monthly_salaries_last10) == 10
    average = sum(monthly_salaries_last10) / 10
    limb_c = 0.5 * average * completed_years
    limb_b = max(0, notified_limit - exempt_used_before)   # lifetime aggregate
    exempt = min(received, limb_b, limb_c)
    return {"average": average, "limb_c": limb_c,
            "exempt": exempt, "taxable": received - exempt}

LIMIT = 10**9   # placeholder: insert the current notified limit yourself
print(gratuity_exempt(250000, [60000]*6 + [66000]*4, 7, LIMIT))
print(gratuity_exempt(250000, [60000]*2 + [66000]*8, 8, LIMIT))

The placeholder limit is deliberately huge so that limb (b) never binds in the demonstration. Replace it with the current notified figure before relying on the result.

The lifetime aggregate most people forget

Section 19(2)(a) makes the notified limit apply across all your gratuity exemptions, not per employer. If you claimed an exemption on an earlier job's gratuity, subtract it from the limit before applying limb (b). The function takes that as exempt_used_before. Keep the settlement statements from every past employer; you will need them to know the figure.

Where the taxable part goes

The ₹31,600 above is salary for tax year 2026-27, and TDS on it is part of your employer's section 392 deduction. Whether it attracts tax depends on the rest of your income. Under the new regime, total income up to ₹12,00,000 is covered by the section 156 rebate. Line it up against the other lines of your exit statement using the full and final settlement audit. Leave encashment on exit has its own four-way test, worked in leave encashment tax.

Averages, date windows and minimum-of-three tests are small analysis exercises, of the kind practised in Python on our Data Science programme.