What Is the Expected Maximum of Two Exponential Variables?

Assume $X$ and $Y$ are independent exponential random variables with rate parameters $\lambda_1>0$ and $\lambda_2>0$. The expected value of their maximum is

When the rates are equal to $\lambda$, this becomes $3/(2\lambda)$. The independence assumption matters: knowing only the two marginal exponential distributions is not enough to determine the expected maximum.

Derive the expectation from the survival probability

For any nonnegative random variable $Z$, its expectation can be written as the integral of its survival probability:

Set $Z=\max(X,Y)$. The event $Z\le t$ means that both $X\le t$ and $Y\le t$. Independence lets us multiply those probabilities. For an exponential variable with rate $\lambda$, $\mathbb{P}(X>t)=e^{-\lambda t}$ for $t\ge 0$, so

Integrate each term:

This method works because the maximum is nonnegative and its survival probability is easy to express using the joint probability that both variables are small.

Example with different rates

Suppose $X$ and $Y$ are independent, with $X\sim\operatorname{Exp}(2)$ and $Y\sim\operatorname{Exp}(3)$, where the numbers are rates. Their individual means are $1/2$ and $1/3$. The expected maximum is

The answer is less than the sum of the means, $5/6$, because the maximum is not the sum. It is greater than either individual mean in this example, as expected: $\max(X,Y)$ is at least as large as each variable for every outcome.

Equal rates and the scale convention

If both variables have rate $\lambda$, substitute $\lambda_1=\lambda_2=\lambda$:

Some books and software describe an exponential distribution using a scale or mean parameter $\theta=1/\lambda$ instead of a rate. In that notation, two independent exponential variables with the same scale have expected maximum $3\theta/2$. Check which convention is being used before substituting a parameter: confusing rate and scale reverses the answer’s units and magnitude.

Why independence matters

The formula uses independence when it writes $\mathbb{P}(X\le t,Y\le t)$ as a product. If $X$ and $Y$ have the same exponential marginal distribution but are identical rather than independent, then $\max(X,Y)=X$ and its expectation is $1/\lambda$, not $3/(2\lambda)$. Thus, without information about how the variables depend on each other, their marginal distributions do not fix the expected maximum.

Common mistakes

• Using the sum formula. For independent exponential variables, the minimum has an exponential distribution with rate $\lambda_1+\lambda_2$. The maximum does not; its expectation is not simply $1/(\lambda_1+\lambda_2)$.
• Dropping the subtraction term. Adding the two means counts outcomes where both variables are large in a way that the maximum does not. The term $1/(\lambda_1+\lambda_2)$ corrects for this overlap.
• Mixing up rate and scale. With rates, the mean is $1/\lambda$. With scales, the mean is the scale itself.
• Assuming independence without checking. If the variables are dependent, use their joint distribution or another description of their dependence; the independent-variable formula need not apply.