How does the P/E ratio actually work?

The price-to-earnings ratio is the most quoted number in all of investing, and also the most misread. Let's build it from the ground up so you own it completely.

Two ways to write the same ratio

P/E divides price by earnings. You can compute it two ways and get the identical answer:

• Per share: share price ÷ earnings per share (EPS). A $150 share with $6 of EPS gives a P/E of 25.
• Whole company: market capitalisation ÷ net income. A $150 billion company earning $6 billion also gives 25.

Both are correct because market cap is just share price times share count, and net income is just EPS times share count — the share count cancels out. Use whichever the data in front of you makes easy.

What the number is telling you

A P/E of 25 has a plain-English meaning: at today's price and today's profit, you are paying 25 years of current earnings for the whole business. Flip it upside down and you get the earnings yield — 1 ÷ 25 = 4%, the profit the business generates each year as a percentage of its price. That flip is a handy sanity check, and it lets you compare a stock's earnings yield against, say, a bond yield.

Why the number is usually "high"

Beginners often see a P/E of 25 and flinch — 25 years sounds like forever. But a growing company's earnings are not frozen. The market is paying for future earnings it expects to be larger than today's. A company expected to double its profit over a few years can carry a P/E of 25 that quietly becomes an effective 12.5 once that growth arrives. This is exactly why a raw P/E means little without context — and why the next unit brings in growth explicitly.

When P/E simply breaks

P/E has a hard limit: it needs positive earnings. A company losing money has no meaningful P/E — the ratio goes negative or undefined, and data providers usually show a dash. Fast-growing firms that reinvest everything, or cyclical firms in a bad year, routinely have no usable P/E. That is not a glitch; it is a signal that you need a different tool, which is what P/S and the enterprise-value multiples are for.

In the data

Here is a loss-making company, Rivian, on every P/E-shaped figure a data provider publishes:

Earnings per share still carries the loss (−2.59 on 29 September 2026). The headline P/E is blank, which is honest: there is no P/E for a company that earns nothing. But the trailing P/E in the valuation block reads 0, a number where there should be none. A missing P/E and a low P/E are different things, and a ranking that treats the blank or the zero as a low multiple puts every loss-maker at the cheap end.

The per-share arithmetic below fails silently for some listings. A price is quoted in the currency of the exchange; the statements are filed in the company's reporting currency; the two can differ. TSMC trades in New York in dollars while its statements are in Taiwan dollars; HSBC is quoted in London in pence while its statements are in US dollars (both read 29 September 2026). Divide one by the other and the P/E is a number with no meaning. Check the quote currency and the reporting currency before dividing. When they differ, use the published multiples, which are built on one basis, or convert at the rate on the statement date.

Try it now

The ratio and the number underneath it, published side by side:

1. Note the P/E and the earnings per share.
2. Multiply EPS by the P/E, then compare the product with the share price:

You should land near one of the two rows, and on 28 September 2026 the product matched the previous close to the cent. Watching the ratio reassemble makes it yours. 3. Flip the P/E upside down for the earnings yield: 1 ÷ P/E, as a percentage. That is the profit the business generates each year against its price, and it is the form you can hold next to a bond yield. 4. Now go back to the loss-maker under "In the data". Say which row still carries the loss and which has gone blank, then find the P/E-shaped row that reports a number anyway, and say what a screener sorting on it would do with the company.