ERA for Relief Pitchers

The common assumption is that a reliever's ERA is comparable to a starter's and that a low number always means a good pitcher. It is not, and it does not. A reliever ERA in the low twos over 15 innings tells you almost nothing reliable. FanGraphs notes that reliever ERA is unreliable below 30 innings and recommends 50 innings for a meaningful comparison. The same formula, (Earned Runs × 9) ÷ Innings Pitched, applies to every pitcher, but the context for a reliever is fundamentally different: short outings, inherited runners, and high-stakes situations make the raw number deceptive.

If you evaluate a reliever for fantasy or analysis, know what the ERA number hides and which stats actually predict future performance.

The ERA formula is identical for all pitchers. A reliever who allows 2 earned runs in 4 innings has a 4.50 ERA, exactly the same as a starter would. The problem is sample size. A starter typically pitches 100 or more innings per season, giving the ERA time to stabilize. A reliever may pitch 50-70 innings in the same period, and many throw far fewer. A single bad outing of 1.2 innings with 4 earned runs can spike a reliever's ERA from 2.00 to 4.50. That 2.00 ERA over 10 innings is not a reliable signal of skill; it is a noise event waiting to be corrected.

The Official Baseball Rules, Rule 9.02, defines innings pitched in thirds: 5.2 IP means 5 and 2/3 innings, or 5.667. Decimal confusion, treating 5.2 as 5.2 instead of 5.667, can misstate ERA by about 0.10 to 0.30 runs. For a reliever with a small sample, that error can change a 3.50 ERA to a 3.80, which changes the evaluation entirely. Always convert partial innings correctly before calculating.

If a reliever enters with two runners on base and both score, those runs are not charged to the reliever. MLB Rule 9.16(c) is explicit: runs allowed are charged to the pitcher who was in the game when the runner crossed the plate, not the pitcher who allowed the runner to reach base. This is the inherited-runner rule, and it is the single biggest reason a reliever's ERA can look better than their actual performance. A reliever who inherits a bases-empty situation faces no penalty. One who inherits runners already on base gets a clean ERA no matter what the previous pitcher left behind.

This also works in reverse. A reliever who enters and walks a batter, leaving runners on base, passes those runners to the next pitcher. The reliever is charged only for runs that score while they are pitching. The common error is to assume that a reliever who allows few runs must be elite. That assumption ignores the possibility that the reliever faced fewer threatening situations or inherited favorable game states. In fantasy leagues, evaluating reliever ERA without checking inherited-runner contexts is a known failure mode.

League-average ERA varies by season and role. For 2025, the MLB average was 4.28 (source: Baseball-Reference). The 2025 average for MLB-affiliated minor leagues was 4.85 (source: Baseball-Reference). These numbers replace the unsourced ranges from older guides that claimed a 3.00-5.00 range for starters and 2.00-4.00 for relievers. Those bands were not tied to any season or level and mislead anyone trying to compare across leagues or eras.

A 4.00 ERA in 2023 was slightly above average, not elite. A reliever with a 3.00 ERA in 2025 is roughly 30% better than the MLB average, but that number means something different for a reliever than for a starter. The starter faces batters from the beginning of the game, often when the order is strongest. The reliever often enters in lower-leverage situations or inherits runners that inflate their ERA risk. The two roles produce different ERA distributions even when skill levels are identical. The claim that relievers have lower ERAs because they pitch fewer innings is incorrect; the real reason is that they often enter with inherited runners, which inflates their ERA relative to starters who begin clean innings. The gap is about 0.30 to 0.80 runs.

Because reliever ERA is noisy and context-dependent, use stats that isolate pitcher control or adjust for situation. FIP (Fielding Independent Pitching) estimates ERA from strikeouts, walks, and home runs only, ignoring team defense. FanGraphs publishes FIP, xFIP, and SIERA; SIERA is the most predictive of future ERA for sample sizes over 50 innings. For a reliever with fewer than 30 innings, even SIERA should be treated with caution.

K-BB% (strikeouts and walks per nine innings) measures the two outcomes a pitcher fully controls. A high K/9 stabilizes ERA faster. BB/9 inflates ERA directly. WPA (Wins Probability Added) accounts for game leverage, a closer who enters with the game tied and gives up a run does more damage than one who enters with a five-run lead, and WPA captures that. Inherited runners scored % tracks the share of runners a reliever leaves on base that later score. A low inherited runners scored % means the reliever strands runners, which is a skill; a high % means the reliever allows runs to score after leaving the game. FanGraphs library and Baseball-Reference glossary define these metrics.

Take a reliever who has pitched 18 innings across 15 appearances with a 2.50 ERA. That means they have allowed 5 earned runs (2.50 = (5 × 9) ÷ 18). Now they enter a game, face three batters, and allow 4 earned runs in 1.1 innings (1.333 innings). The new totals: 18 + 1.333 = 19.333 innings; 5 + 4 = 9 earned runs. New ERA = (9 × 9) ÷ 19.333 = 81 ÷ 19.333 = 4.19. The ERA moved from 2.50 to 4.19 in one outing.

This is the failure case. A fantasy player who drafted this reliever based on the 2.50 ERA has just lost 1.69 runs of assumed value in one night. FanGraphs notes that reliever ERA is unreliable below 30 innings, and this example shows why: the sample is too small to absorb a single bad event. The same pitcher over 150 innings would have a far more stable ERA. The solution is to use FIP or SIERA for evaluation before the sample reaches 30 innings, and even then to treat the number as provisional.