Search for average serve speed and you will find neat tables: so many mph for a 3.5, so many for a 4.0. We looked for the research behind those tables and found none. There is solid data for professionals, some for juniors, a little for beginners, and almost nothing for club levels.
Below are only the numbers we could trace to a study, with how each was measured. The studies used different methods, so treat them as separate reference points, not one ladder.
What professionals average on first and second serves
The clearest recent figures come from a 2023 study in Biology of Sport. The researchers recorded 630 first serves and 315 second serves from 21 male professionals at the 2019 Barcelona Open, an ATP 500 event on red clay, using the Foxtenn system. Two radar guns at the centre of the baseline, 2 m (6 ft 7 in) high, recorded the peak speed of each serve.
| Serve | Average speed | Typical spread (1 SD) | Each player's fastest, averaged |
|---|---|---|---|
| First serve | 173.6 km/h (107.9 mph) | 18.0 km/h (11.2 mph) | 201.3 km/h (125.1 mph) |
| Second serve | 144.7 km/h (89.9 mph) | 13.1 km/h (8.1 mph) | 163.8 km/h (101.8 mph) |
Second serves were about 17% slower than first serves, as professionals trade pace for spin and margin. And the average first serve sat well below each player's fastest: the big numbers on a stadium speed board are the top of the range, not a typical serve.
This is one tournament, on clay, with men only. We did not find a comparable open, peer-reviewed sample of women's tour serves, so we give no women's average rather than borrow a figure we could not check.
What research says about juniors and college players
A 2019 PLOS One study tested 128 Spanish juniors chosen by their regional federations as the most talented players in each region. Each player hit 30 maximal flat serves, with a radar gun 3 m (about 10 ft) behind the server at roughly contact height. The score was the average of each player's eight fastest serves.
| Group | Boys | Girls |
|---|---|---|
| Under 13 (average age 12.6) | 126.4 km/h (78.6 mph) | 122.0 km/h (75.8 mph) |
| Under 15 (average age 14.6) | 149.1 km/h (92.6 mph) | 132.4 km/h (82.3 mph) |
These are best-effort flat serves in a test, not match averages, which would probably be lower once second serves count. For college players we found only small samples, mostly conference abstracts, several without average speeds, so we have no college figure we would trust enough to print.
What is known, and not known, about club players and NTRP levels
We found no study that measured serve speed with radar across NTRP levels, or in a large sample of club players. The closest peer-reviewed figure is a 2005 French study in Medicine & Science in Sports & Exercise. It split 32 players into beginner, intermediate and elite groups and measured the maximum ball velocity of power serves by radar. The elite group reached 169.4 km/h (105.3 mph) and the beginners 107.2 km/h (66.6 mph). The abstract we read does not give the intermediate figure.
The per-level tables online come from coaching estimates, informal surveys and self-reports. A self-reported number may come from a radar held at an angle, one lucky serve or a guess, so we do not reproduce them. The USTA level descriptions we have read describe the serve in words, such as placement, spin, power and dependability, and none gives a speed. Your level comes from match results, as our NTRP guide and 4.0 player guide explain.
- ATP men, first serves, match average174 km/h (108 mph)
- Elite group, maximum serve (2005 study)169 km/h (105 mph)
- Top junior boys under 15, best serves149 km/h (93 mph)
- ATP men, second serves, match average145 km/h (90 mph)
- Top junior girls under 15, best serves132 km/h (82 mph)
- Top junior boys under 13, best serves126 km/h (79 mph)
- Top junior girls under 13, best serves122 km/h (76 mph)
- Beginner group, maximum serve (2005 study)107 km/h (67 mph)
Why speed matters less than you might think
Even at the top, speed is only part of the story. A 2017 study in the Journal of Sports Sciences used Hawk-Eye data on 25,680 first serves by 151 men at the Australian Open. Hitting aces depended more on accuracy than on speed: serves aimed more than about 6 degrees away from the returner and landing within about 15 cm (6 inches) of a service box line were the most likely to be aces.
Speed also costs margin. A faster flat serve has a smaller window over the net, as our guide to why serves go long works out from the court's measurements. So a few extra km/h can lower your first-serve percentage and put more pressure on your second serve. Check whether a slightly slower first serve wins you more points with our guide to first serve percentage, and make sure your second serve holds up before chasing pace.
Why radar guns and apps give different numbers
- Peak versus average speed. A radar gun reports the fastest speed it sees, which is close to the racket, just after contact. Air resistance slows the ball all the way to the bounce, so it reaches the returner well below the number shown. The professional figures above are peak readings.
- Where the radar sits. A radar reads the true speed only when the ball travels straight toward or away from it. Held off to the side, it reads lower, and the wider the angle, the bigger the error.
- One serve or many. A single best serve, the average of your eight best, and the average of every serve in a match are three different numbers. The junior study used the eight best of 30.
- Video and phone apps. Apps that estimate speed from video depend on frame rate, camera position and how they find contact and bounce. We have not tested them, so treat their numbers as rough unless the app explains its method.
How to measure your own serve roughly
- Set up the radar behind you. Place it in line with the path of your serve, at about contact height. The junior study put it 3 m (about 10 ft) behind the server, at about 2.2 m (7 ft 3 in).
- Warm up, then serve 20 flat first serves to the same target. Write down each reading and whether the serve landed in.
- Use the average of your best few serves that landed in, not the single fastest one. A serve that misses tells you little.
- Repeat with the same setup each time, so you compare your serve with itself. Your own trend means more than any table.
No radar? Film side-on in slow motion and time the ball from contact to bounce. A serve down the middle that lands near the service line travels roughly 18 m (60 feet) along the court, so 0.45 seconds means an average of about 40 m/s, or 144 km/h (89 mph). That average reads lower than a radar's peak. At 30 frames per second, one frame either way moves the result by about 10 km/h (6 mph), so use your phone's highest slow-motion setting. Our filming guide shows where to put the phone.
What actually adds serve speed
A 2006 biomechanics review by Bruce Elliott in the British Journal of Sports Medicine estimates how much each arm movement contributes to racket speed at impact in a power serve: about 40% from upper-arm internal rotation at the shoulder, about 30% from hand flexion, about 15% from horizontal flexion of the upper arm and about 10% from the shoulder. Pronation of the forearm contributes about 5%, and its main job is setting the racket face. These figures leave out the legs.
The legs and trunk still matter. The same review describes a vigorous leg drive as what stretches and pre-tenses the shoulder's internal rotators before the swing. In the 2005 French study, elite players pushed harder into the ground than beginners (a peak vertical force of 2.12 against 1.68 times body weight) and hit the ball higher. Both findings come from small groups, and the second shows a difference between levels, not proof that pushing harder will make you faster.
- Keep the swing continuous. Elliott cites a study in which upper-arm internal rotation was about 20% faster with no pause than with a 1.5-second pause, so avoid stopping at the bottom of the knee bend.
- Accept what you cannot train. In the ATP study, body height and mass explained 22% to 35% of the variation in players' fastest serves.
- Build up gradually. The review notes that shoulder and elbow loading rose with serve speed, and that players with effective knee bend and extension had lower loading.
Our guide to improving your serve turns these points into video checks and drills, and our kick serve guide covers the spin side.
TennisAI, our video analysis app, does not measure serve speed. It scores your serve motion from an uploaded clip and gives feedback on timing, contact point, body position and balance. The score is an automated estimate that varies with camera angle and lighting, so film from the same spot each time, and use a radar for how fast you serve.
Frequently asked questions
What is a good serve speed for a club player?
No study we found sets one. The closest research figure is a beginner group at about 107 km/h (67 mph) maximum in a 2005 study. Compare your serve with your own earlier readings, and judge it by first-serve percentage and points won.
How fast do professional tennis players serve on average?
In one study of 21 men at a 2019 ATP clay event, first serves averaged 173.6 km/h (107.9 mph) and second serves 144.7 km/h (89.9 mph), measured as peak speed by radar.
What is the fastest tennis serve ever recorded?
The 2023 Biology of Sport study notes a 263 km/h (163 mph) serve recorded in an official men's match, and adds that the ATP does not officially recognize serve speed records.
Does TennisAI measure serve speed?
No. It gives a serve score and feedback on your serve motion from an uploaded video, so use a radar gun if you want a number for how fast you serve.
See this in your own game
Upload a practice or match video to TennisAI and get feedback on your forehand, backhand and serve.
Sources
- Baiget, Corbi and López, Biology of Sport (2023): Influence of anthropometric, ball impact and landing location parameters on serve velocity in elite tennis competition
- Fernandez-Fernandez et al., PLOS One (2019): Age and sex-related upper body performance differences in competitive young tennis players
- Girard, Micallef and Millet, Medicine & Science in Sports & Exercise (2005): Lower-limb activity during the power serve in tennis: effects of performance level
- Whiteside and Reid, Journal of Sports Sciences (2017): Spatial characteristics of professional tennis serves with implications for serving aces
- Elliott, British Journal of Sports Medicine (2006): Biomechanics and tennis
This article is general information about tennis training and does not replace a coach or medical advice. Stop and see a professional if something hurts. TennisAI is made by Pace, the publisher of this blog. TennisAI is not affiliated with USTA, UTR Sports, the ITF or any official tennis association. Rating systems change, so check the official sources for current rules.



